(navigation image)
Home American Libraries | Canadian Libraries | Universal Library | Community Texts | Project Gutenberg | Children's Library | Biodiversity Heritage Library | Additional Collections
Search: Advanced Search
Anonymous User (login or join us)
Upload
See other formats

Full text of "Electronic computers and management control"

I 



UNIVERSITY 
OF FLORIDA 
LIBRARIES 




ENGINEERING SCIENCES 
LIBRARY 



ELECTRONIC COMPUTERS and 
MANAGEMENT CONTROL 



Digitized by the Internet Archive 
in 2013 



http://archive.org/details/electroniccompuOOkozm 



ELECTRONIC COMPUTERS and 
MANAGEMENT CONTROL 



GEORGE KOZMETSKY 

Litton Industries, Beverly Hills, Calif. 

PAUL KIRCHER 

Associate Professor of Accounting, School of 
Business Administration, University of California 

Los Angeles, Calif. 



McGRAW-HILL BOOK COMPANY, INC 

New York Toronto London 

1956 



ELECTRONIC COMPUTERS AND MANAGEMENT CONTROL 

Copyright © 1956 by the McGraw-Hill Book Company, Inc. Printed in the 
United States of America. All rights reserved. This book, or parts thereof, may 
not be reproduced in any form without permission of the publishers. 

Library of Congress Catalog Card Number: 55-12105 



Preface 



This book is written primarily for the business executive. Our pur- 
pose is to explain certain new developments which may have a greater 
influence on the management of enterprise than any other single group 
of events have had since the first industrial revolution. 

These developments include electronic computers and other new 
equipment for recording, processing, and communicating information. 
Of equal importance, though not as well publicized, are new methods 
for scientific analysis of business data. 

Together these developments are making possible automation of the 
office and of the factory. 

The book does not require that the reader have technical training. 
Therefore it should be helpful to anyone who is interested in coming 
changes in business and society. Automation will have a major im- 
pact on our way of life. To understand why, it is necessary to have 
some knowledge of the way electronic computers operate and of the 
scientific approach to management problems. 

The book describes the fundamental characteristics of electronic 
systems and the basic concepts of the scientific methods of analysis. 
It shows how they will influence management planning and control. 
It describes dozens of actual applications and gives briefer references 
to many more. For readers interested in further details, the appen- 
dixes and the selected bibliography provide additional sources of in- 
formation. 

During the three years that this book was in preparation, the au- 
thors have been aided by so many people that it is impossible to list 
them all. Each of the many references to actual experiences or to 
particular equipment means that someone ( often a group ) has helped 
us. In particular, however, we should like to thank W. W. Cooper of 
Carnegie Institute of Technology, Richard Canning of Canning, Sisson 
Associates and of the Management Science Project at UCLA, Irving 

v 



vi Preface 

Lieberman and Jack Thorne of Litton Industries, Erwin Tomash of 

Sperry Rand, and Don Wall of International Business Machines. 

Perhaps the greatest contribution has been the patience of our wives 
and children for three long years of lost week ends, evenings, and 
holidays. 

George Kozmetsky 
Paul Kircher 



Contents 



Preface v 

1. Introduction: Why Be Interested in Computers? 1 

2. What Is an Electronic Computer? 7 

3. Survey of Electronic Methods of Data Processing 32 

4. Studies and Applications of Electronic Systems 49 

5. Administrative Problems Experienced in Introduc- 

ing Computer Systems 97 

6. Management and the Scientific Approach 116 

7. Management Planning and Control 139 

8. Programming, Scheduling, and Feedback 148 

9. Integrated Business Systems 169 

10. Automation and Scientific Computation 190 

11. Role of the Executive in Selection of an Electronic 

System 212 

12. The Challenge to the Executive 229 

Bibliography 245 

Appendix 1. The Language of the Computer 251 

Appendix 2. Programming 258 

Appendix 3. Electronic Data-processing 

Equipment 265 

Appendix 4. A Mathematical Model for an Inte- 
grated Data Systems 275 

Index 291 



vii 



CHAPTER ONE 



Introduction: Why Be Interested in Computers? 



The development of the electronic computer has created new op- 
portunities for management planning and control. Automatic handling 
of information can become more rapid, more efficient, and often less 
costly than ever before. 

As our economy has become more complex, and as corporations 
have grown in size, in variety of product, and in extent of market, 
the need for improved systems of planning and control, and data 
processing, has become increasingly evident. Executives have rec- 
ognized this need. They have purchased or leased office machines, 
analyzed procedures, and encouraged attempts to raise clerical ef- 
ficiency. They have studied reporting systems. They have attempted 
to relate the gathering and processing of information to the decision 
needs of the organization. 

Nevertheless, the information-gathering and communication proc- 
ess has remained among the less efficient and most unsatisfactory of 
business functions. This situation has led executives to a considera- 
tion of the possible use of new inventions and developments in the 
field of electronic processing of information. 

An electronic system can: 

1. Increase transmission, processing, and reproduction speeds. 

2. Reduce the need for manpower. 

3. Reduce storage-space requirements. 

4. Automatically handle intermediate steps in data processing, thus 
giving more flexibility in preparation of a variety of reports, 
while at the same time increasing accuracy. 

Reduction of clerical cost has been a major factor in favor of using 
electronic systems. As it has become increasingly evident that a 

1 



2 Electronic Computers and Management Control 

computer can pay for itself, more and more companies have placed 
orders for computers or have initiated studies leading to expected 
installations. 

However, recent experience has shown that the major impact of 
these new systems will come in a field of even more significance 
than reduction of clerical cost. The proper use of electronic systems, 
together with application of new methods of analysis, can revolu- 
tionize management planning and control. 

Improvements in the system for planning and control will: 

1. Provide more useful information concerning operations, for use 
in making better decisions. 

2. Enable management to relate the procurement, production, and 
sales programs for each division in a way that results in the 
largest profits for the company as a whole. 

3. Help management to make the best allocation of financial and 
other resources. 

4. Eliminate certain management functions and strengthen others. 

A few of the first companies to link several plants were U.S. Steel, 
du Pont, General Electric, Standard Oil of New Jersey, and SKF In- 
dustries. Special groups are assigned to the study of the potential 
benefits of the new developments. Electronic systems are installed 
to link various plants to central computers. 

Don Mitchell, chairman of the board of Sylvania Electric, announced 
that his company's reason for building a data-processing and com- 
munication center was to serve an integrated area and not any single 
plant. He also said that cost savings were not the consideration. His 
competitors had only two years within which to match his anticipated 
system for providing management with complete daily information. 

Scientific Developments. Electronic computers are only one of 
a number of scientific developments of immediate concern to the 
executive. Among these developments are new methods of recording 
information, of storing it, and of printing it. In addition, a series 
of inventions have appeared which are summed up in the term "auto- 
mation." * Since so much information originates in the factory, and 

* The word "automation" is believed to have been used first by D. S. Harder, 
Ford Company vice-president, to describe a system for automatic handling of 
parts between progressive stages in production. About the same time, John 
Diebold also used the word to mean both automatic handling and the process 
of making things automatically. The latter usage has become more general. 



Introduction: Why Be Interested in Computers? 3 

since electronics is of direct importance in factory automation, it 
is expected that computers eventually will be used for controlling 
production processes as well as for handling clerical information. 

In both office and factory, and for certain types of engineering and 
other research as well, there also has been a development of cer- 
tain new scientific methods of analysis. These methods use tools 
which are primarily mathematical in nature, although not necessarily 
so. One of the names given to this development is "operation re- 
search." Since the new concepts emphasize the application of sci- 
entific methods to business problems, a somewhat broader name also 
given to this new field is "the management sciences." In any case, 
these new concepts are of direct concern to the businessman. 

Need for Timely Executive Attention. There are a number of 
reasons why decisions concerning the use of the new equipment and 
the new methods will have to be made by top executives, and soon. 
The attempt to apply scientific methods to the problems of general 
management has frequently been made before, but with little success. 
Failure came for a number of reasons, but a major factor was the lack 
of equipment for rapidly gathering and processing data and the lack 
of scientific tools with which to attack business problems. Now, 
the equipment is available, many of the required scientific tools are 
available, and more are being developed. 

In order to compete effectively in the future, a company may have 
to make extensive use of the new equipment and new analysis 
methods. This may require an extensive reconstruction of its sys- 
tem of management planning and control, substantial revisions of 
all data-handling processes, and perhaps a realignment of the or- 
ganization structure. It may, for example, mean that a company 
should reverse a present trend toward decentralization because con- 
trol data can be assembled and analyzed in a central office, from 
across the country, as easily and as quickly as they now are sent to 
an office across the street from the factory. Instructions can be re- 
turned as easily and quickly. The additional expense for communica- 
tion may well be insignificant compared with savings. 

On the other hand, the tools may be used to further decentralization 
of decision making. Distant managers now can be provided with full 
information. 

A new approach to the decision-making process itself may become 
necessary. Teams of specialists may make scientific analyses of data 



4 Electronic Computers and Management Control 

and report to management in ways that will be considerably different 
from and more effective than those employed at present. 

Top-management Considerations. If for no other reason, these 
developments will command top attention because of their cost. 
Electronic computers are expensive at present, although they will be- 
come less costly as time goes by. Automation, system studies, train- 
ing of personnel, etc., will add substantially to the cost. 

Moreover, personnel problems will be acute. Large numbers of 
factory and clerical workers can be displaced. Workers with spe- 
cial skills for machine maintenance, systems study, and advanced 
data analysis must be retrained or hired. 

Decisions will have to be made as to whether the machines and 
methods should be introduced gradually, in a sort of evolutionary 
process, or whether the whole matter is so revolutionary that no 
halfway approach will do. Further decisions concern the type of 
equipment. A rather surprising number of new manufacturing firms 
are entering the field. Executives will have to choose from several 
suppliers and may find it best to assemble a unit from components 
produced by several different firms. 

Similarly, advanced methods of data analysis are not immediately 
adaptable to most business problems. Careful examination of specific 
situations must be made before the methods can be used. 

Management has the responsibility to administer this transition 
with the least harmful impact on society. At the same time managers 
have an important role to play in securing the full advantages of 
the new developments for their own companies. They have to eval- 
uate wholly new concepts of doing business. While they will not 
have to understand all the details of new tools and methods, they 
will be responsible for decisions as to where and when these are 
to be used. • 

Computers and automation will require long-term commitments of 
capital. Since these investments can be justified only if they con- 
tribute to earnings over an extended period, a forecast of such con- 
tributions must be made. Frequently this estimate can be little 
more than an intelligent guess, in view of all the factors which bear 
on the decision. 

This problem leads to a consideration of the environment as a 
whole, the framework within which the company must try to operate. 
Businessmen have become increasingly aware that their results are 



Introduction: Why Be Interested in Computers? 5 

influenced to a major degree by factors over which they have little 
or no control. One of these is the result being achieved by all other 
units in the economy. 

Another major factor, in a sense controlled and yet almost equally 
difficult to predict, is governmental action. Through its regulations, 
spending, borrowing, financial controls, psychological impacts, etc., 
government has a tremendous influence on business. 

The evaluation of these factors is obviously a difficult task. For- 
tunately, many of the same analysis methods which are being pro- 
posed and used for other aspects of the business are of direct use in 
studying these problems. 

Need for Technical Understanding by Management. Executives 
cannot achieve their goals of meeting competition and of serving the 
many interests in the business and society unless they are alert to 
new developments which will enable them to administer their busi- 
ness more effectively. But though this is admitted, it does not follow 
that developments in electronic computers, scientific management, and 
automation are the best answer to their specific problems. The new 
machines and methods offer no panacea. For many companies they 
probably will not offer much assistance, at least within the immediate 
future. 

What can the individual executive do? 

He can study these developments sufficiently so that he will be 
able to assess their significance to his particular business. This is a 
responsibility that he cannot delegate. He can learn in general what 
the machines will do, how the methods of data analysis operate, what 
automation means— not in great detail, but sufficiently so that he 
can assess recommendations as to when and where they can be put 
into use in his business. 

Does this mean the administrator with technical training will have 
an advantage? 

It does, for even to communicate effectively with the experts in these 
new fields a certain degree of technical understanding is required. 

Should tomorrow's administrators be chosen solely from men with 
technical training? 

Not necessarily. For years, as technically trained men have moved 
into executive positions, they have tried to bring their past skills to 
bear on the administrative problems of their companies. As a rule, 
however, they found that administrative problems were more com- 



6 Electronic Computers and Management Control 

plex than they expected. Therefore, like the nontechnical admin- 
istrators, they have had to rely also on personal experience and the 
"art" of management. This will continue to be so. 

Summary. The executive need not become an expert in electronics 
or mathematics. As with other technical areas of business, an execu- 
tive can learn to deal with the problems involved by acquainting 
himself with general aspects— with capabilities and limitations of the 
systems and methods. For the details, he can rely on experts. His 
task is to select the experts and to administer their efforts so as to pro- 
duce maximum benefits for his company, his customers, his employees, 
and society. 

The purpose of this book is to assist the executive by describing 
the major characteristics of electronic systems and new methods of 
analysis as these are related to management planning and control. 
The book attempts to provide the background required in order that 
the executive may obtain the economic benefits offered by these new 
developments. 



CHAPTER TWO 



What Is an Electronic Computer? 



The use of electronic computers for business purposes is a develop- 
ment of the 1950s. Before this decade a few computers existed, 
but they were used solely for scientific computations. Since 1950, 
however, the growth of interest in electronics among businessmen 
has been remarkable. Articles in popular magazines have heralded 
the new "electronic brains." Professional journals, and meetings of 
professional societies, have discussed the potential of this new equip- 
ment. Orders for hundreds of computers of all sizes have been 
placed by businessmen. 

A variety of claims have been made— that these new devices will 
handle all the data processing of a large business at a great reduction 
in cost; that they can read print, make computations, prepare state- 
ments and payroll checks, and so on. Sometimes statements are 
heard that the new systems will replace management itself. 

What are these electronic computers? How do they work? Who 
is using them? How are they being used? In what ways does the 
business operation change when they are introduced? 

Businessmen have found it difficult to get a simple answer to these 
questions. The electronic computers they have examined, and the 
ones which have been described as soon to be available in the future, 
differ so much in abilities, costs, and other characteristics that the 
machines hardly seem to belong to the same family. 

Like the blind men describing the elephant, it is easier for the 
experts to talk in terms of the parts of the machines rather than of 
the whole. They speak of input devices, storage, logical elements, 
etc. These technical aspects seldom have a direct significance for 
the executive. 

Electronic computers use electrical charges or magnetic marks to 

7 



8 Electronic Computers and Management Control 

represent business information, to compute with it, to store the data, 
and to drive a printer which produces numerical or alphabetical char- 
acters in familiar forms on reports. However, when one looks to see 
how the electrical charges are activated, he finds that they may come 
from data-processing devices currently employed, like punched cards. 
Or they may originate from some new form, like magnetic tapes. 
Some of the data may be taken from keyboards that have numerical or 
alphabetical keys, just as those currently used for ordinary typewriters 
or adding machines. 

Furthermore, many of the operations of an electronic computer are 
not strictly "computation." The machines are also used to rearrange 
data, or to do logical processing, e.g., to select cost variances which are 
more than a certain range off standard, and to print out a report for 
special attention by management. 

A major reason why it is so difficult to define "electronic computers"' 
is the fact that the field is in a stage of transition. There is no single 
type of machine, or even a single type of process, which dominates the 
field. It is necessary to speak of electronic computers, particularly of 
those used for handling business data, in terms of "electronic systems." 
By electronic systems is meant a combination of various devices, rather 
than a specific machine or method. 

General-purpose Computers. Electronic systems can be classified 
into two general types : general-purpose computers and special-purpose 
machines. 

General-purpose computers, as their name implies, are electronic 
machines that do all types of arithmetical computations— add, subtract, 
multiply, divide, square root. In addition, general-purpose computers 
are capable of performing a large number of other operations. They 
can store data, compare items, prepare reports, and so on. As a result, 
they are very versatile. For example, they can do the payroll, handle 
accounts receivable and billings, do cost accounting, production sched- 
uling, and budgeting. 

A general-purpose computer can be regarded as a data-processing 
center, one that is completely integrated and able to perform all of 
the various functions of data processing: receiving information, con- 
verting information, storing data, sorting data, collating data, comput- 
ing data, transmitting data, and putting the data into a usable form, 
as readable printed output. 

A general-purpose computer, from a machine point of view, is a 



What Is an Electronic Computer? 9 

combination of devices. These devices are linked in a way which per- 
mits the machine to perform operations by manipulation and transmis- 
sion of electrical charges or magnetic marks. From the functional 
point of view, a general-purpose computer is a system which permits 
the user to process data for a large number of purposes, all within 
a centralized grouping of devices, which are primarily electrical in 
nature. 

General-purpose computers can be further classified according to 
size and to the speed of computation. General-purpose computers 
may be either "large-scale" or "medium-sized." Computers which can 
handle a variety of problems and are extremely fast in arithmetical 
computation are generally referred to as large-scale general-purpose 
computers. Their rentals range from $15,000 to $50,000 a month. 
Computers which are smaller in size and slower in computational 
speed, but still able to handle a variety of problems, are referred to 
as medium-sized general-purpose computers. They usually rent at 
$2,000 to $6,000 a month. As yet there are no "small-size" general- 
purpose machines that sell for $1,000 to $25,000, although design engi- 
neers are nearing this goal. 

Special-purpose Computers. Special-purpose electronic computers 
are limited either as to the type of computations they can make or as 
to the functions they can perform. For example, a special-purpose 
computer can be devised which only adds and subtracts. A machine 
designed to handle only customer billings, such as Bell Telephone's 
new automatic billing machine, can be termed a special-purpose com- 
puter. 

In contrast to the general-purpose computer, the special-purpose 
computer is designed to handle separate aspects of data handling. 
A special-purpose computer might handle the recording of the number 
of calls but might not be able to compute the monthly telephone bill. 
The addition of more devices may enable the system to compute the 
monthly bill but still not handle payroll. 

If the system were redesigned, however, so that the central computer 
could also make payroll computations, it might then be a general- 
purpose computer. Yet much of the "billing-special" system still might 
be retained. The same input system, for example, might be used. 

Special-purpose computers also can be classified by their size, large 
or medium, and by their speed. 

The limited variety of functions which it can perform, both mathe- 



10 Electronic Computers and Management Control 

matical computations and business data-processing functions, makes a 
machine special-purpose; but for certain operations the special com- 
puter may outperform a large-scale general-purpose computer. 

MISCONCEPTIONS 

As a result of the publicity given to electronics in recent years, many 
people seem to believe that computers are actually electronic brains. 
Electronic computers are indeed marvelous inventions, but they cannot 
replace the judgment of the human brain. A computer can do only 
what it is ordered to do. The machine can carry out a number of in- 
structions in programmed sequence, rapidly and automatically (and 
not talk back). But if a single one of the necessary instructions is 
omitted because it was not originally prepared by a human, the ma- 
chine must be ordered to stop or its results will be useless or misleading. 

Although electronic systems will revolutionize many business sys- 
tems and procedures, help pull together fragmented responsibilities, 
provide a means for integration of decentralized operations, and elimi- 
nate many clerical and middle-management functions, they cannot 
provide a completely automatic administration. They cannot formu- 
late and define the purpose of an organization. This will remain a 
function of human management. Computers are not brains. 

Cost. As another result of the publicity given to the first scientific 
computers, many persons believe that these new devices are only of 
interest to firms which have several million dollars with which to 
experiment. Since most businessmen are accustomed to paying for 
office equipment in terms of tens of thousands of dollars, or less, they 
may conclude that this is all far beyond their capabilities. 

It is true that the general-purpose large-scale computers, which have 
tremendous capacities, are expensive. But already there are many 
types of smaller and cheaper electronic systems, and more are being 
developed. Some very inexpensive models (under $5,000) are de- 
signed for special purposes. Under development are smaller general- 
purpose systems which will cost considerably less than present large- 
scale computers; some will cost less than $30,000. Yet the savings they 
can offer are significant. Moreover, most of the machines need not be 
bought outright; they can be leased. It is not even necessary that 
companies have the machines in their own offices. Service centers are 
being established where machines can be rented for a few hours a 



What Is an Electronic Computer? 11 

day. Since the larger machines operate so rapidly, a few hours are 
sufficient to handle a large volume of work. 

Types of Systems. It is extremely important that the businessman 
understand that there are many kinds of electronic equipment. Some 
can replace an entire tabulating-machine installation. Some can re- 
place only a hand-posted file. Some can be used only to record the 
count of items. Others are only calculators. Some are sorters. Others 
can store data on an electronic or magnetic medium, such as a tape, or 
drum, and can reprint it when desired. All of these abilities can be 
had at different speeds for different costs. 

Generally speaking, the more flexible and rapid the machine, the 
more expensive. Large-scale general-purpose systems, in which all 
types of abilities must be present, obviously cost the most. 

Another misconception is that the machines are all so similar that 
the choice is largely "whether a company should go ahead now, not 
which system it should buy." But the individual businessman also has 
the problem of choosing a machine, or several machines, to form the 
data-processing system which best meets his particular needs. This 
decision, of course, will be reached in the same way that any capital 
investment decision is made. If substantial savings can be forecast, 
the task will be easier, but the details of the expected costs must be 
available. Proposals to invest in electronic systems must compete with 
each other, as well as with the alternative uses of capital which are 
always before management. The value of one new system as against 
another should be judged not only by the benefits and the amount of 
the expected saving, but also by how certain it is that such results 
can be achieved. 

Boards of directors or top managers often like to see how a machine 
or system has worked elsewhere before they try it themselves. But 
while it is possible to wait and learn from others' experience, this is 
not as desirable an approach to electronic systems as it may be in some 
other types of cost study. In general, the reason for this is that each 
system must be tailored to individual company requirements. 

The result of several studies has indicated that wherever more than 
a hundred clerical workers are engaged in computing, sorting, tran- 
scribing, or storage of data, a general-purpose electronic system merits 
consideration. Special-purpose information, such as recording an in- 
ventory situation, can be obtained by using special systems if the 
present expenditure or the added value of the information is of the 



12 Electronic Computers and Management Control 

order of about $15,000 to $20,000 a year. Systems have already been 
accepted that call for a pay-off period of up to 5 years, because of other 
advantages in speed and accuracy. 

A number of companies are already buying computer time on a 
service basis. Rates of $100 to $300 per hour are being quoted. Even- 
tually, many more companies will be able to reduce costs through 
judicious use of this type of service. Rental use of computers is now 
a widely accepted practice for engineering calculations by companies 
whose own facilities are overloaded. 

Preparation for Installation. There is more to be considered than 
the cost of the machine and the systems study. Personnel with many 
new types of ability must be hired or trained. A detailed description 
of the system type of data transferred, management needs, etc., must 
be prepared. The "engineering specifications" of the paths and rates 
of flow must be established. 

This preparation was overlooked by one company. It was one of the 
first to order a large-scale computer, but it was forced to request 
postponement of the installation. 

The computer must be told what it is to do in great detail and in a 
"language" code that it can understand. Of necessity the people who 
deal directly with the machine must be specially trained. Since most 
machines have heretofore been working on mathematical problems, the 
only groups who have had much experience in feeding data to machines 
have been the electronics engineers and the mathematicians. The first 
companies to venture into this field hired such men to advise them and 
to lead the way. 

Studies and subsequent experiences have shown, however, that for 
data processing it is easier to train a businessman in computer abilities 
and requirements than it is to train a technical expert in business-sys- 
tem requirements. This means that usually it is advisable to take a 
capable man or a group from a present systems and procedures force 
and to train them in the use of computers. Usually this has been 
accomplished by sending the men to the schools now operated by the 
major manufacturers of computers. Schools such as Harvard, M.I.T., 
Wayne, and U.C.L.A. also have begun to offer computer courses. 

Mathematicians take longer to learn the intricacies of a particular 
business system than systems men take to learn what they need to 
know about computers. In spite of this, however, men with some skill 
in mathematics can make a significant contribution toward the more 



What Is an Electronic Computer? 13 

efficient use of a large-scale computer. Mathematical training is useful 
in formulating and stating the problem to be solved. A "program of 
instructions" for a computer— the detailed list of steps it follows in 
processing a group of data— must be logically complete, as must a 
mathematical solution. Moreover, as the company advances in its 
use of new methods of analysis, the presence of several mathematicians 
becomes all the more desirable. 

Preparation of each "program" for a data-processing routine on an 
electronic computer takes more time than is sometimes believed. It 
is often a matter of several months rather than weeks. Each step must 
be studied by expert workers, and all possible exceptions considered. 
However, once the program is established, it can be stored in the com- 
puter itself, or in media rapidly insertable in the computer, so that it 
can be used over and over with a minimum of further study. In 
some machines part of the program can be changed, if it is desirable 
to do so, without losing all the benefit of the work already done. 

Speed. The speed at which present electronic systems operate is 
another point commonly misunderstood. Certain aspects of their 
operational speeds are phenomenal. Some computers can perform 
thousands of arithmetical operations a second. But in some systems 
the original recording of the data is done no more rapidly than it is 
done at present. For example, most payroll applications of electronics 
now contemplated still use timecards, from which the data will be 
transferred to other media for entry to the computer. The day will 
come when counters and meters and other instrumentation will auto- 
matically record all sorts of information and transfer it directly to the 
computer. But this is still in the future for most applications. 

Often the high speeds of computation which can be achieved are 
greater than are needed. For example, processing of payrolls in a 
large company now requires a number of days, partly because of the 
amount of computation involved. Electronic systems can reduce the 
time for computation to hours. Whether they reduce it to 4 or to 6 
hours will rarely be significant. Speed of computation is not the only 
criterion for judging the effectiveness of these systems. Speed, more- 
over, adds to the complexity and cost of the electronic system. 

Equally important is the speed with which data can be introduced 
and taken from the computer. At present several electronic systems 
print out information only six to ten times faster than the rate achieved 
by printers of the type in common use in punched-card installations. 



14 Electronic Computers and Management Control 

Several very high-speed systems are under development, but there is a 
question whether one such printer will be better than several slower 
ones. In any case, substantial improvement over present methods 
is still fairly expensive, and much remains to be done. 

In many business situations data are recorded in random sequence. 
Because large volumes of data are usually stored in serial fashion, 
on magnetic tapes, sorting speeds are important. Since sorting seldom 
was required for mathematical and scientific calculations, early com- 
puters were not designed to sort rapidly. A number of solutions to this 
problem have recently been developed, but equipment now offered 
is not always economical. The problem of rapid access to large vol- 
umes of stored information is receiving considerable attention from 
computer manufacturers, and will be discussed later in more detail. 

In many types of application there is need for better balancing of the 
various capabilities of the systems before they can attain the speeds 
which are popularly believed to be available. Much work is now 
under way on these problems. Study is still needed also to determine 
the best combination of characteristics in order to handle business 
needs in an optimum fashion. A number of experts who have been 
studying the requirements of their particular companies are still far 
from satisfied with the equipment now on the market. 

Obsolescence. This leads to another popular misconception— that 
computers now being offered will shortly be obsolete. Changes are 
expected along two fronts. There will be improvements in the over-all 
design, and in various components and parts. 

It is true that the first models of any machines are inevitably experi- 
mental. Each manufacturer that builds a machine incorporates de- 
signs which reflect a particular viewpoint. In addition, machines 
offered commercially as a rule use parts which have been tried and 
tested, not the newer devices which offer considerable hope for im- 
provement but which are not yet reliable. 

As business needs are increasingly understood, as designs are com- 
pared and shaken down, and as parts are improved, present machines 
will tend to become obsolete. But before deciding to wait for better 
equipment, the buyer should consider several factors which tend to 
offset the fear of obsolescence. Tremendous sums already have been 
invested in electronic computers, and much of the first experimentation 
has been accomplished. Secondly, in many installations the immediate 
savings and increase in operating effectiveness will be sufficient so that 



What Is an Electronic Computer? 15 

the systems can pay off the original investment long before economic 
obsolescence is likely to occur. 

In addition, experience gained with the electronic system will reduce 
the expense of installing and using any new and improved system. 
This factor should be of real significance. Moreover, several expected 
improvements, such as the introduction of high-speed printers operat- 
ing at over 1,000 lines per minute, can be tied to existing computers, 
so that only a portion of a system will be replaced. 

If the user feels the danger of rapid obsolescence is too great, the 
equipment can be leased. Most manufacturers of computers make 
them available under reasonable leasing arrangements. 

Much publicity has been given to diodes, transistors, and magnetic 
cores. Diodes are units that will pass current only in one direction. 
They are now used in some computers in large quantities— as many as 
1,000 to 5,000. Engineers tend to overspecify in order to protect 
against machine failure. Even so, many types of diodes cost an average 
of only about $.75 to $2.00 each, so that if the cost were reduced sub- 
stantially, the difference in the total cost of the initial system probably 
would not be too significant. Of course, more reliable diodes reduce 
maintenance costs. 

Transistors are units which perform functions similar to vacuum 
tubes, but use less power and last longer. Reliable transistors are 
becoming available in quantity. Most computers will be equipped 
with them some day, it is presumed. Companies already are designing 
their computers to take advantage of transistors as they become more 
available. They also design sections which are only plugged into the 
original equipment, so that they can be replaced when better sections 
are available. Use of transistors will reduce costs, weight, power re- 
quirements, and need for air conditioning, but probably will not alter 
the data-processing capabilities of the machines significantly. 

Magnetic cores, described in Chapter 3, offer a method of storing 
data where it can be reached in a few millionths of a second. Cores 
are more reliable than some other presently used devices, but they are 
not revolutionary in the sense that by using them the machines will 
obtain new types of ability. 

The general conclusion is that these developments offer hope for 
improvement, but they are not likely to be a decisive factor in reducing 
the cost of most types of equipment during the next few years. 

However, the electronic industry is one of rapid development. En- 



16 Electronic Computers and Management Control 

tirely new concepts of logical design are being studied. Some day 
these may well enable manufacturers to produce general-purpose ma- 
chines for significantly less cost than at present. The business user 
should find it advantageous to keep informed of these developments. 

Accuracy and Dependability. As with any business machines, man- 
agers are much concerned with the accuracy and dependability of the 
electronic system. Because of early difficulties with electronic com- 
puters put together by experimental groups, it is often mistakenly be- 
lieved that computers are not too dependable, that they break down 
frequently. 

Machines may use as many as several thousand tubes and circuits, 
and though the percentage of failure is low, instances of failure may 
be fairly frequent— perhaps several a week. However, the engineers 
of manufacturing companies are well aware of the problem and have 
incorporated a number of solutions in their computers. 

Computers are now being built with plug-in or "modular" units. 
Testing equipment often is made part of the computer itself. Tubes 
usually weaken before they fail, and this weakening often can be de- 
tected. As a result, troubles can be located as part of routine mainte- 
nance. The offending circuits can be removed and replaced in a 
matter of minutes. Stores of these parts are maintained at the com- 
puter or are available from the manufacturer. A few hours of pre- 
ventive maintenance at regular intervals minimizes unscheduled down 
time. 

Scientific computers have been improved considerably in recent 
periods, so that along with scheduled preventive maintenance of about 
10 per cent of the time, unscheduled breakdowns have averaged less 
than 10 per cent of the time. Several large commercially built com- 
puters have not been down unexpectedly for more than 4 hours at a 
time since installation, although they are working around the clock. 

Many businessmen are worried lest their records be lost in an elec- 
tronic system. Some types of record are lost if power fails. If the data 
are in these types of storage devices, the sources of information should 
always be recorded elsewhere, in types of storage that do not fail if the 
power goes off. Then the information always can be reproduced. 

Data on magnetic tapes and drums do not disappear if the power 
fails. Still, since the data are in the form of magnetic charges, they 
may be erased from tapes or drums inadvertently by stray electric 
shocks. To protect against loss from such accidents a complete repro- 



What Is an Electronic Computer? 17 

duction of all information in the system can be run off periodically and 
transferred to safe storage. If any disaster occurs, the company can 
back up and start over. Running off the duplication ( which is placed 
on magnetic tapes similar to those used for sound recording) is not 
complicated, time-consuming, or overly expensive. 

Engineers have made considerable efforts to achieve the accuracy 
required for accounting and auditing standards. A number of checks 
have been devised to test the operation of commercial computers. As 
a result, in some ways electronic systems already are more accurate 
than any other type of data-processing system. Some can make long 
series of numerical calculations with no errors that the machines them- 
selves fail to catch. Of course, if wrong information is introduced, it 
will be repeated, but even here certain automatic checks can be em- 
ployed. For example, the machine can be instructed to refuse to ac- 
cept nonexistent inventory code numbers that a human tries to enter. 

On the other hand, the effect of some types of errors in the system 
may be serious. Fewer human observers scan the data as it is proc- 
essed. Electronic systems usually are less fragmented than are decen- 
tralized clerical operations, so that the effect of an error has somewhat 
greater chance of becoming compounded. However, practical safe- 
guards against this also can be devised, using intermittent print-outs, 
or using the high speed of the system to cross-check and repeat. 

In general, computers have shown themselves to be highly accurate, 
making considerably fewer mistakes for the same amount of work than 
do other systems. They have made some mistakes, but workers believe 
that most of them have been detected. 

Summary of General Concepts. Electronic systems are great inven- 
tions. They now offer considerable benefits to business users, even 
though they are not brains endowed with judgment. They are fast, 
accurate, and dependable. Present machines will not become obsolete 
overnight. The cost is not unreasonable in view of the service and 
savings which they can provide to business. The benefits are there for 
those who are willing to learn how to use them. 

THE GENERAL-PURPOSE ELECTRONIC COMPUTER 

To this point we have been discussing the general concept of elec- 
tronic data-processing systems. As was indicated, there are a variety 
of electronic devices available for data processing. Some are large- 



18 Electronic Computers and Management Control 

scale general-purpose computers, some are smaller scale, others are 
special-purpose electronic devices. In order to help the reader to 
distinguish sharply between a general-purpose computer and the less 
extensive electronic devices, this section will attempt to describe in gen- 
eral terms how a large-scale digital computer works. 

The word "digital" refers to the fact that the machines considered 
for accounting and other business data-processing needs usually are of 
a type which records, computes, and reports in terms of absolute 
digits, like a desk adding machine. Some electronic machines are 
"analog"; like a thermometer or speedometer, they can indicate only 
approximate figures. The analog computer has many uses in engineer- 
ing, where it can simulate physical conditions, in utility power grids, 
flight conditions of airplanes, and so on. There have been suggestions 
concerning possible applications to economics and business. This will 
be investigated later; meanwhile, the system discussed is "digital." 

The basic concepts involved in the description of digital computers 
can be grouped as follows: 

1. The components. These are the functional units which are 
grouped to form a general-purpose electronic computer. Several 
mechanical devices may be grouped as a "component." 

2. The functions. These are the operations carried on by each of 
the components; the type of work each performs. 

3. The devices and media. The devices are the mechanical units 
which carry out each of the functions. Usually the devices are 
categorized in terms of the "media" they use. "Media," such as 
tapes, or punched cards, hold the data while they are being proc- 
essed or stored. The distinction between a "device" and a "com- 
ponent" is somewhat arbitrary, but here the larger units will be 
called "components." 

4. The operating characteristics. The different capacities and other 
characteristics of the devices, such as their speeds of operation, 
bulk, etc., are called "operating characteristics." 

A large-scale digital computer has the following major components: 
input, processor, and output. The organization of these components 
can be compared with the operation of a desk calculator. The key- 
board of the calculator is the "input," the mechanical counters perform 
the arithmetic and storage functions of the "processor" while the opera- 
tional keys provide control, and the dials are the "output." 



What Is an Electronic Computer? 19 

In the electronic system all these units are electrically connected so 
that data can be transferred between them. Figure 1, a simplified 
diagram, gives an indication of this linkage. 

Table 1 shows the interrelationships of these concepts in a more elab- 
orate fashion. Most large-scale computers have all of the major com- 
ponents listed. They usually offer different devices to carry out each 
function, which gives them varying operating characteristics. 



INPUT- 



-*• PROCESSOR 



OUTPUT 



MANUAL OR 

AUTOMATIC 

INPUT 



STORAGE 



INPUT 




ARITHMETIC 

AND 

LOGIC 



SWITCHING AND CONTROL 



OUTPUT 



OBSERVABLE 
OUTPUT 



Figure 1 



How a Computer Works— Input. As in any other manner of data 
processing, the first operation is to record the data. If this original rec- 
ord is not made in a form which the computer can handle and read, 
the record must be converted onto media which can be used as input. 
Special devices are available to handle this operation. Input to the 
computer is usually thought of as comprising the original recording 
of the data, the conversion to other media if necessary, and the actual 
transfer of the data into the processing component. 

Large-scale computers usually are not directly linked to the point 
of original recording of data. Source documents of the type now used 
still must be prepared. The data from these documents usually must 
be transcribed, either simultaneously or later, on media such as mag- 
netic tapes, or paper punched tapes, or punched cards. This must be 
done in a manner which can be read by the computer. In other words, 
the data must be placed on the media in a form which is the "lan- 
guages-symbol system of the computer. (See Appendix 1 for discus- 
sion of the language of the computer. ) 

Print readers, which will translate properly printed material, are 
now coming on the market. Other devices prepare and read input 
in the form of coded dots, etc. Once the data have been transcribed 



20 



Electronic Computers and Management Control 





o 
o 




o 


£? 


o 


1 1 


8 


sq 


o 



"08 fl 
T3 a 



St 



6B.3 H- 1 

a S 3 

S S O 

-d . o 

2b 



M OS 

u a 3 

.2 0* 00 

CD „ CO +3 

-d Si? © 



-d d B 
a o.2 



a® 

CD -fl 



ft (0 

o 

-si 

83 J 
TJ 



§.2 j 03 



m S to 

0! « " 



o d 

■e 2 



0) m 

cc S 

>> 9 



3 03 
O T3 



8> 



=aj 



•a **» 
§ s g 

05 a 



a 3 



s a 



< cs3 



is 

4) "43 
53 o 



T3 ft 



CD O 






3 CD 

IS 



CD O 

as 



&2 



d CD 55 

M 2J CD 

C S ° 

"S rS 2 

d "S fe 

hH P ft 



if 

I s 

H ft 



2"! 

2ft 






8 £ 

Ph d 



What Is an Electronic Computer? 21 

onto the input media, the source documents are never again required, 
except possibly for auditing purposes. In effect, the data are ready 
to go through a production line, each stage of which can be controlled 
automatically. 

The manner in which the data flows down the production line is 
governed by the control element of the processor. 

Control. The control device determines whether the computer is 
to read input, or to add or subtract, or where to store results, etc. In 
order for the control element to operate, it must be given a prepared 
list of instructions. People who deal with computers call this set of 
instructions a "program." 

The program is stored in the computer in magnetic or electronic 
form, just as is the information to be processed. The program is 
entered via the regular input. When in the computer, the program 
governs the operation of the circuits. It tells the mechanism when and 
how to perform each individual step in the routine which is being per- 
formed. 

The program must instruct the mechanism which item to read off 
the input media, where to transfer it, what to do with it ( for example, 
to multiply it, etc. ) , and where to put the result. If any single instruc- 
tion in the program is not correct— and some programs have thousands 
of steps— the machine will stop or give incorrect results. 

Logic and Arithmetic. The processor takes instructions from the 
program. It can transfer data from one part of the computer to an- 
other. It can compare numbers and choose its next operation by jump- 
ing to different steps in the program automatically as a result of the 
comparison. One small example: while posting requisitions of inven- 
tory, it can initiate a purchase order if the inventory balance falls 
below the stored reorder point. Another useful ability of the logic 
device is to let the control component know when the computer has 
reached the end of a routine; it will automatically start the next 
sequence or stop the machine. 

The control tells the arithmetic unit which mathematical operation 
to perform, at what step in the process. The arithmetic element can 
add, subtract, multiply, and divide, shift decimal points, round off 
numbers, etc., and temporarily store results. In general, these opera- 
tions can be performed at the rate of several hundred to several 
thousand computations per second, depending on the size of the 
machine. 



22 Electronic Computers and Management Control 

Storage. Storage is the holding of data for later use in the operation 
of a computer. The input stores data in forms that can be brought 
to the computer. It transfers the data to temporary storage locations 
in the computer or directly to the main storage device. 

The major types of internal storage devices usually are classified 
according to their speed-of-access time. This "access" speed is the 
length of time required to locate a storage address and to place an 
item in, or to take an item from a storage unit and to make it available 
to another part of the computer. Some types of storage are "serial"; 
for example, the items on a tape reel can only be reached at high speed 
in their recorded order. The length of time it takes to reach any item 
of information, no matter where placed in the storage, is spoken of as 
the "random-access" time. Obviously, random-access time is of great 
importance for certain purposes. For example, with high-speed ran- 
dom-access storage there is no need to sort input. Generally speaking, 
storage units with high-speed random access cost more per unit of 
information that they can store. As a result, only a limited amount of 
such rapid-access type of storage has been built into present computers. 
High-speed storage devices include electrostatic tubes, acoustic delay 
lines, and magnetic cores. 

Moderate random-access speed, to greater volumes of stored infor- 
mation, is available economically on magnetic drums and magnetic 
tapes. To give a rough indication of speeds of access to stored data, 
some magnetic drums can provide information in a few thousandths 
of a second, while electrostatic storage or magnetic cores can provide 
data in a few millionths of a second. 

Because the arithmetic and logical units operate so rapidly, speed of 
access to stored data is of prime importance for efficient operation of 
the present types of computers. Since most computers refer to their 
stored program of instructions at each step of the operation, the over- 
all speed of operation can be no faster than the highest access speed 
to such storage. 

Output. The results of computation can emerge either in a form 
suitable for record storage in the system, or in a printed form. The 
former type of output is often placed on magnetic tapes similar to 
those used for input purposes. The tapes can be stored as they are, or 
the data can be transferred automatically to punched cards at the rate 
of more than a hundred cards a minute. 

A variety of printers are in use. Some were developed long before 
computers were perfected. Examples are electric typewriters (six to 



What Is an Electronic Computer? 23 

ten characters a second ) , and line-at-a-time printers such as those used 
in punched-card systems (100 to 150 lines a minute). High-speed 
printers, under continuing development, print up to a thousand lines 
(100 to 130 characters each) a minute from either tapes or cards. 
Photo devices record several thousand characters a second. 

The program of the computer can control the format of the printed 
product, so that reports used by business can be prepared automatically 
in any form desired which is compatible with automatic printing. 
Most systems provide separate printers, so the main computer is not 
tied up during the printing process. 

ILLUSTRATION OF THE OPERATION OF A COMPUTER SYSTEM 

As can be seen, the operation of an electronic computer designed for 
processing business data is not easy to understand. However, it is 
possible to follow the major stages of the operation by means of a 
simplified illustration, such as the one which follows. 

Suppose a company wishes to place its inventory accounting on a 
computer. Assume that the problem is merely ( 1 ) to record an open- 
ing balance, (2) to record the subsequent purchases and sales and (3) 
to compute current balances each day, including the ending balance, 
and (4) to make this information available as required. 

Suppose the company records increases in inventory on the basis of 
invoices after confirmation by receiving room reports. It receives 
notices of sales from the sales department as sales orders; the assump- 
tion is made that all orders taken result in shipments. 

Conversion from present equipment to the electronic system is a 
problem which will be discussed later in the book. For present pur- 
poses, let us assume that the company in effect is starting an entirely 
new operation, with no inventory on hand. 

Input. Each item of inventory must be given a code identification, 
as in many present systems. When an invoice arrives, plus a receiv- 
ing report showing, for example, that 11 units of item No. 356 have 
arrived, unit cost $5.00, the information is sent to a girl who types it 
on a regular typewriter keyboard. The machine she uses, however, 
not only types the information on a standard form, but also automati- 
cally translates it into a series of electrical signals, which are placed as 
magnetic charges upon the surface of a tape coated with material that 
will store the impulses. ( Or, the medium used might be paper tape, 
in which case the machine would perforate the tape as the girl typed. ) 



24 Electronic Computers and Management Control 

After a suitable amount of information has accumulated on the tape, 
it is taken from the typist and placed upon a device which feeds the 
tape past a reading head, much as sound tape is fed past a reading 
head on a sound-tape recorder and player. 

We now have the "input" ready for action. 

Processor. However, before the computer can operate, the machine 
must be "instructed" as to what it shall do with the information as the 
data are read by the reading head. These instructions, of which there 
will be many even for a process as simple as the one we are describ- 
ing, are the "program" of the machine. An analogy to a computer 
program is the wired plugboard of a regular punched-card machine, 
which routes the impulses read off the card, and thus achieves the de- 
sired arithmetical operations and transfer of other information. But 
the analogy is not exact. A computer program is a list of commands 
which are stored in, can be modified by, and are carried out by other 
parts of the mechanism. They are not a "way routing," such as is the 
plugboard. 

Assume that we store a list of instructions in the machine. The first 
of these instructions will be, in effect, "Read the number that is under 
the reading head." Thus the code No. 356, identifying the type of in- 
ventory item, will be read as it comes under the reading head. 

Next, a subsequent instruction (or part of the first instruction- 
machines vary as to how many parts there are to each instruction) 
will tell the computer, in effect, "This number is the same as that given 
to a storage position (address) in the computer's storage. Connect 
your circuits with that storage position and read out the total number 
of items shown to be on hand at present." This instruction is analogous 
to telling a clerk to go to a file drawer ( No. 356 ) and to copy the cur- 
rent balance, in units, from the inventory card. In this case, it has 
been assumed that this current balance is zero. 

When the amount of items on hand— zero— is determined, it is trans- 
ferred to a temporary storage position in the arithmetic part of the 
computer, where it is available for computation. This temporary stor- 
age position is often called a "register." A register is an electronic 
circuit that can store an item of information, such as a number or an 
alphabetic character, and make it rapidly available ( in a few millionths 
of a second ) when again needed. All general-purpose computers have 
such registers. 

Now the program tells the computer to read the next bit of infor- 



What Is an Electronic Computer? 25 

mation on the incoming tape. This is the fact that "eleven" new items 
have been received. The program tells the circuitry to add this amount 
to the "zero" stored in the register, and place the sum back in address 
position No. 356 in the storage. The equivalent manual operation 
would be to have the file clerk transport the number to another clerk, 
who added the two items on a comptometer and gave the answer back 
to the file clerk, who placed it on the card and returned it to the file. 

The system described would be sufficient if all that were wanted was 
a straight unit count of the inventory balance. There are actual elec- 
tronic systems now in operation which do about this and no more. 

However, if information concerning dollar values is also wanted, 
the electronic system can handle this too. The machine must be in- 
structed to hold the "eleven," which it has just read from the tape, in 
the register. Thus it is still available for computation, even though it 
has also been added to the previous unit balance. Then the machine 
is instructed to read the unit price from the tape, $5.00, to multiply the 
"11" by "5," and to add the total, $55.00, to another part of address posi- 
tion No. 356. Of course this means that the storage position must have 
been previously subdivided, say into No. 356.10 and No. 356.11, etc., 
so that the machine will not confuse the units count with the dollar 
total. 

Sales can be handled in the same way, except that the machine 
would be instructed to subtract. The typist would prepare the tape 
containing the sales information in a fashion similar to the tape for 
the purchases. It is even possible to have purchases and sales mixed 
together on the same input tape; the sales would be given an identify- 
ing code that differentiated them from the purchases, so that the ap- 
propriate computation could be made. 

The system described above assumes that the storage can be searched 
in random fashion, as a file clerk can search a set of files— that is, it 
can go to any one of the "drawers" at random. "File computers" using 
sets of drums or discs can do this. Other systems of electronic or mag- 
netic storage now in use, while large enough for the use of tables, or 
listings of a few thousand items, are not large enough to handle in this 
way the many thousands of items of inventory customarily found in 
business. 

This means that a system of storage on tapes may have to be used, 
which in turn may mean that the items must be brought to the com- 
puter in a sorted order. As item No. 356 comes in on the purchases 



26 Electronic Computers and Management Control 

or sales "input," the No. 356 storage position will be made available 
from the tape storage, and the operation can proceed as before. This 
is like having all the incoming items sorted in numerical order, so the 
human clerk can walk through the files in a straightforward fashion, 
without retracing his steps, inserting new computations in the files as 
he goes. 

Output. Since the information concerning the inventory is stored in 
the form of electrical signals, or magnetic charges, on various types 
of storage media, it is necessary to have some means of making it 
available for humans to read. This is accomplished by adding a 
printer to the system. This device can operate from the computer 
and, in effect, can reverse the process that was used to record the data 
as they were typed from the invoices and other papers. 

The above system is highly simplified. A great virtue of these ma- 
chines is that they can do so many things automatically. For example, 
as part of the instructions, the machine might be told that when a sale 
is made from storage address No. 356 the machine should examine the 
lowered balance and compare it with a minimum balance for that item, 
also stored in the system. If the sale takes the balance below this mini- 
mum figure, the machine can note that fact in a separate storage loca- 
tion, which can be used to print out a list of such items for the atten- 
tion of the purchasing department. Or, if the name, address, etc., of 
the usual supplier and the standard order amount are stored in the 
computer, it can print out a purchase order automatically. 

The process of comparison is carried out by the logical circuits of 
the machine. In essence, the machine can take a number from one 
location, another number from another location (the latter may even 
be a computed number, as in the example ) , and subtract one from the 
other. If the first is the larger, or the two are equal, the machine will 
do one thing ( that is, go on to another item ) . If the second is larger, 
so the difference between the two amounts is a negative number, the 
machine will do a different operation (that is, list the item as having 
fallen below the minimum balance, then go on to another item ) . 

To illustrate a still more complicated procedure, suppose that the 
managers would like to use the "moving average" method of inventory 
calculation. This procedure has always been too time-consuming for 
widespread acceptance. But it is a simple matter for the computer. 
It can be instructed to add a new purchase to the old ( both the number 
of units and the dollar figures ) , to divide the new dollar total by the 



What Is an Electronic Computer? 27 

new unit total, to calculate the new average unit price, and to use this 
new average unit price for any subsequent sales reductions of inventory. 
The arithmetic and logical components of the computer are so effi- 
cient that manufacturers have advised potential users to think in terms 
of operations even more complicated than this. For example, it is not 
only feasible but efficient and economic for a computer to work through 
a group of insurance policies, calculating anew the dividend available 
on each. In other words, it can do the computation for each policy 
separately according to a complicated actuarial formula involving 
dozens of factors. The computation would only take a fraction of a 
second, and is more efficient than storing a large table of precomputed 
dividends. 

INTERNAL OPERATION OF A COMPUTER 

There is some debate as to the amount of information which an 
executive should have as to the internal operation of a computer. 
Some argue that, as with a telephone or an automobile, it is not neces- 
sary for a user to know much of the physical attributes of the system 
in order to use it efficiently. Others point out that even these relatively 
simple devices perform more efficiently for the user who knows why 
he should wait for the dial tone or keep oil in the crankcase. 

There is nothing weird or fantastic about the operation of an elec- 
tronic computer, although it is a very complicated mechanism. The 
more an executive understands about the principles of operation, the 
better will he be equipped to assess the recommendations and sug- 
gestions which he will receive from his staff advisers concerning the 
installation and use of electronic equipment. 

A brief introduction to the internal operation of the computer is 
presented in Appendixes 1 and 2, under the headings "The Language 
of the Computer" and "Programming." 

Electronic Storage. One of the simplest, yet most disturbing, attri- 
butes of the computer is its ability to store information in a form which 
cannot be seen or felt. Yet it is more permanently and safely recorded 
than it would be on the more familiar medium, paper. 

The basis for this record is the fact that material can be magnetized 
or that electric currents can flow in two different physical states. 
Magnetism can be placed directionally, i.e., either the north or south 
"pole" can be "up." Properly magnetized material appears to maintain 



28 Electronic Computers and Management Control 

this state indefinitely, and will withstand conditions that would destroy 
paper records. 

Since there are these two discernible states, they can serve as sym- 
bols. If "north" up is considered to be "zero," then "south" up would 
be considered to be "one." On this basis it is possible to record com- 
binations of "zeros" and "ones." These can be coded to be the equiva- 
lent of any combination of decimal numbers, of alphabetic characters, 
of punctuations, of placings in report forms, and so on. The symbol 
system is not efficient— it takes a good many "0"-"l" symbols to represent 
our more usual numbers and characters, but the machines can transfer 
these symbols at the rate of 186,000 miles a second and store them 
several hundred to the inch along a tape and in several channels across 
a tape. One computer design calls for a 31-channel tape, 2,800 feet 
long. One reel, which would about fit in a case for a portable type- 
writer, will hold the equivalent of over 400,000 punched cards. 

Electronic Processing. Another major attribute is the ability of the 
computer to transfer these "0" and "1" symbols in such a way as to 
perform addition, subtraction, or the other processing steps which have 
previously been described. This ability is derived from the physical 
characteristics of certain electronic devices. One of these is the fa- 
miliar vacuum tube, or its cousin, the transistor. 

A vacuum tube or transistor has three principal elements. Lines 
lead to each of these. A current coming in line No. 1 can come out 
(often amplified) on line No. 3 only if the proper signal is present on 
line No. 2. Thus the line No. 2 acts as a switch. By proper combina- 
tion of circuits, it is possible to have these switches allow currents to 
pass only when the computer program signals that it is desired. 

The designer of the system builds his machine so that a certain signal 
will always cause the circuits to add incoming signals from specific 
locations. Another signal causes the circuits to place the result in a 
specific location. And so on. 

The designer provides the user with a list of these program signals, 
called "instructions" or "commands." Because these instructions are 
so simple, each one dealing with a very specific detail of an operation, 
it takes a fair-sized number of them to perform even the most simple 
processing of data. Moreover, the same end result can be built up in 
a number of different ways. This means that one program can be 
much more efficient than another. Even though the machine operates 



What Is an Electronic Computer? 29 

very rapidly, if it must perform a routine many thousands of times, the 
differences in time may add up to a costly amount, especially when 
the alternative uses of computer time are considered. 

There are many ways to approach the problem of programming an 
operation. One is to study the situation until the analyst can set up a 
"flow chart" showing the major operations which must be performed. 
Then he studies each of the operations to determine which elements 
must be repeated. An example might be the step which must be taken 
each time a sale is recorded. 

A series which will be repeated is called a "loop." The programmer 
attempts to make the machine perform the loop as efficiently as pos- 
sible, since it may have to be used many times. 

Then he links the loops by providing a transfer command. The ma- 
chine will leave one loop and go to another set of instructions when- 
ever a planned situation occurs, such as a comparison of two numbers 
that results in a negative difference. 

After going through a number of loops, the last transfer may send 
the machine back to the first loop again, using new input, so that the 
whole process can be repeated. In this way the machine can be set 
up to work automatically through a complete record of a day's sales, 
or of the plant payroll, giving each of the like items similar treatment. 

It is evident that skilled programmers are one of the more valuable 
assets a company can possess when installing and operating an elec- 
tronic system. Many firms have spent thousands of dollars per man 
to educate their employees in the art of analysis and programming. 
( This figure includes the salary while learning. ) Many companies are 
contemplating the establishment of analysis and programming teams 
of 10 to 15 men. The larger companies have 30 to 50 and even more. 
Obviously the investment in this aspect of electronics is considerable, 
even apart from the cost of the equipment. 

SUMMARY 

The preceding sections have dealt in a general way with the abilities 
of electronic computers and have explained the importance of a pro- 
gram. 

The fundamental logical principles of a digital computer can be 
summarized as follows: 



30 Electronic Computers and Management Control 

1. Characters (either numbers or letters) can be represented by 
electric signals. In a digital machine, these signals can be used 
to represent the characters precisely, so that the characters can 
be transferred and reproduced as originally recorded. 

2. Arithmetic operations and certain logical operations, such as com- 
parison and choice, can be performed by electronic circuits. 

3. The operations can be controlled automatically by instructions 
stored in the form of electric signals. 

4. The instructions can be coded as alphabetic characters or num- 
bers. 

5. The signals for the characters can be stored electronically or mag- 
netically. 

6. A stored program of instructions can be used to control a sequence 
of arithmetic and logical operations. 

These principles make it possible for business to develop systems 
which have the following general characteristics: 

1. Consolidated files. 

2. Use of predetermined standards, for comparisons. 

3. Simultaneous processing of data for different purposes, eliminat- 
ing intermediate steps. 

4. Provision of exception data as required. 

5. Integration of data-processing and reporting systems. 

Some knowledge of the way computers operate— their "language," 
circuits, components, media, and so on— is necessary for the nonexpert 
who wishes to use electronics for business applications. Understand- 
ing the technical features and limitations will help him in the selection, 
installation, and establishment of realistic schedules. In addition, the 
need for a "team approach" to electronic-system application will be 
appreciated. Members of the current planning and procedures organi- 
zation should be called upon to help the team with their knowledge 
of the company's systems, as well as with their expert skills in various 
areas, but they should not be expected to carry the burden alone. 

The complex nature of electronic computers, as evidenced by their 
complicated circuitry, indicates that large-scale computers will not 
become cheap overnight. But their price will come down as newer 
methods of design appear and as production costs for components and 
for assembly are reduced. 



What Is an Electronic Computer? 31 

A systems and procedures group can do a great deal to prepare for 
future use of a computer, long before the installation is actually made. 
With an understanding of the requirements of electronic systems, they 
can incorporate in their current studies the type of detailed description 
of the operations which must be set down before a program can be 
prepared. They can even prepare a program, test it on a machine 
rented for a few hours or days, establish standard subroutines, and 
so on. 

Programming a computer is painstaking, detailed, and exacting 
work. Unfortunately, the computer is not a self-starting, self-deter- 
mining "brain." The computer cannot analyze and digest data, except 
in so far as its stored program and built-in logical component tell it how 
to do so. Each step of a complicated business procedure must be 
analyzed into the sequence of specific operations that the computer 
can perform. The resulting program must be coded into machine 
language and stored in the computer before it can be used. This 
means that a crew of programmers must be trained as to the require- 
ments of both the computer and the business system. Also that they 
will require time to actually program the system after they have 
learned their job. It cannot be done automatically or mechanically, 
although sometimes subroutines can be reused for more than one 
application. 

On the other hand, the logical operations which the machines can 
perform create opportunities of real importance. Clerical and analyti- 
cal routines which consist of comparison, or arithmetic computation, 
or signaling of maxima or minima, etc., can be performed by the 
computer. The computer will make analyses with tremendous speed 
and with reduced errors and cost, compared with clerical help. The 
proper use of the logical and arithmetical components will help open 
the way for adoption of methods of control and analysis which have 
never been attempted before, because they were obviously not feasible 
without the assistance of a computer or without extended training of 
middle-management personnel. 

The next step in our discussion will be to investigate in more detail 
the operations performed by each of the components. 



CHAPTER THREE 

Survey of Electronic Methods of Data Processing 



Certain fundamental aspects of electronic systems make the systems 
substantially different from any data-processing methods heretofore 
used. (Of course, it should always be remembered that they can 
be linked with older systems, especially with punched cards.) But 
this does not complete the problem posed by the advent of electronics. 
The possibilities offered by these new inventions are so broad and 
varied that there is no single machine on the market which embodies 
all of them. It is difficult to select the equipment best suited to a 
business' needs. 

There are a large number of companies— over two hundred— which 
are interested in the electronic field to the extent that they supply parts 
for the computer systems. There are more than a dozen which are 
building computers. Some have been conducting experiments and are 
still determining what they wish to offer commercially. Others have 
definitely committed themselves to certain types of design, have built 
systems and components, and are offering them for sale. 

Until a prospective user has made a survey of the field, it is difficult 
for him to assess the particular systems which may have come to his 
notice. There is a bewildering array of ideas and "hardware" (ma- 
chines and equipment) which are offered or talked about. Few ex- 
perts are available for advice who are not working for one of the 
manufacturing concerns, although accounting firms and management 
consultants are studying the field intensively so as to be able to aid 
their clients. 

In addition, many businesses already are conducting much of their 
data processing with fair efficiency, or much of it may call for per- 
sonal contact, so that only limited areas may exist for the use of elec- 

32 



Survey of Electronic Methods of Data Processing 33 

tronics. In some cases scheduling of production, or reporting for cen- 
tral purposes, or rapid access to an up-to-date file of information, etc., 
is the major problem which needs solution. For these companies a 
limited use of electronics, embodied in a "special-purpose" system, may 
be economically advantageous. 

Because available electronic equipment is so varied, it is not feasible 
to describe all the devices individually. A study of needs and of re- 
lated electronic possibilities must be undertaken by each business on 
an individual basis. This will require, on the part of the purchaser, 
an understanding of some of the technical aspects of electronic media 
and devices. In general, it is usually possible to think of the processing 
requirements of a company in terms of the amount of raw data which 
must be recorded (input), transmitted and manipulated (processor), 
and the reports and records which must be prepared (output). For 
this reason, a better understanding of the components of an electronic 
system should prove helpful to the businessman. 

INPUT MEDIA AND DEVICES 

As mentioned in the previous chapter, there are three major types 
of input media in use: magnetic tapes, paper tapes, and punched cards. 
In addition, there is direct input from keyboards and from automatic 
devices. 

For a high-speed computer to operate efficiently it must be equipped 
with high-speed input equipment. The machines which offer the 
highest input speeds at present use the magnetic tapes. Magnetic 
tapes can feed data to the processor unit in some systems at the rate of 
over 15,000 characters per second. Several systems operate at about 
10,000 characters a second, which is many times the speed of punched 
cards or paper tapes. 

However, the rate at which data can be recorded by humans on any 
medium, including tapes, is limited by manual typing speeds— about six 
to eight characters a second. For this reason, manufacturers are work- 
ing on systems which will prepare tapes as the data are typed for the 
first time, thus avoiding repetition of this slow and costly bottleneck. 

When the information is on punched cards or paper tape, it can be 
transferred to magnetic tapes. Special devices can accomplish this 
at about 100 to 115 characters per second, the limitation being due to 
the speed of handling the cards or paper tapes. 



34 Electronic Computers and Management Control 

Tabulating Cards. Where a company already has a tabulating-card 
system, it may prefer to continue using a large part of it. In par- 
ticular the company may continue to collect the information, to place 
the information on cards or on paper tapes, to verify it, and so on, be- 
fore entry is made into the electronic system. This will be especially 
true where there is no pressing need for immediate up-to-the-minute 
information. Where the company wishes immediate knowledge of the 
status of an account, or of the location of a car of merchandise, etc., 
direct inputs may be necessary. Nevertheless, punched cards retain 
many of the virtues in an electronic system which have made them 
so widely accepted for mechanical processing. 

A typical punched-card system operates with machines driven by 
electricity, but this does not make it an electronic system. As we have 
shown, in an electronic system the data are transferred, computed, and 
stored in the form of magnetic marks or electric charges. In the older 
punched-card machines computation is done mechanically, much as in 
a desk calculator. The instructions for these machines are fixed by 
wired control panels. They can be changed only by removing one 
plugboard and inserting another or by rewiring the panel. In this way 
they differ considerably from a general-purpose computer, which can 
store a large number of instructions and can change part of the in- 
structions in accordance with other stored instructions. 

Card Programmed Calculators now available might be considered 
an intermediate step between the older mechanical machines and the 
newer electronic computers. 

Magnetic Tapes. Magnetic tapes are useful where the quantity of 
information to be handled is very large. Not only are they one of the 
quickest known ways of entering information in an electronic com- 
puter, but they are one of the best means of storing very large quanti- 
ties of information in the least amount of space and at least expense. 

Magnetic tapes used in electronic systems are similar to those used 
in sound recording. The tape backing may be of plastic, metal, or 
paper. The surface is coated with magnetizable material, a sort of 
paint of iron oxide particles. It is necessary for the coating to be 
complete and even, for a small gap in the coating would lose infor- 
mation, which may be placed as closely as 200 magnetic marks (repre- 
senting binary bits * ) to the inch. For this reason the tapes are not 

* Binary "bit" is the name given to the "0" or "1" which together constitute 
all the symbols used in the binary system of counting. See Appendix 1 for de- 



Survey of Electronic Methods of Data Processing 35 

inexpensive. By the time they have been tested they may cost as much 
as 3 to 5 cents a foot. However, a foot of tape can hold a great deal of 
information, the amount depending on the way the information is 
coded and grouped. And the tapes can be reused many hundreds of 
times, whereas paper tape and punched cards must be replaced by new 
tape or cards. 

A magnetic tape is usually apportioned into a number of "channels," 
which are longitudinal divisions. Generally speaking, each channel 
is read by a separate head in the input device which transfers the 
information to and from the computer. The heads are tiny electro- 
magnets. Current passing through a coil sets up a magnetic field 
in the channel underneath it, on the tape. Or, a magnetized area 
already on the tape, passing rapidly by, induces a pulse of current in 
the coil; the pulse is amplified and taken into the machine. Another 
head usually is provided to erase the mark on the tape when desired. 
A character, comprised of several binary bits, frequently is recorded 
across the tape through use of several channels (as is done with the 
holes in paper tape for teletype). 

Other Media. A drawback to the use of most input devices of 
computers at present is that the information must be transferred 
from the original paper evidence, such as the invoice. Under develop- 
ment now are input devices which will automatically scan printed 
data and activate magnetic tapes. Methods of using specially prepared 
cards or other media are under consideration which will reduce the 
time and cost of transcribing the information. Mark-sensing punched 
cards may speed up the input if they allow for bypassing the key 
punch. Typewriters with special attachments are available for re- 
cording data either on cards or on paper or magnetic tapes at the 
same time that invoices or other source documents are being pre- 
pared; the cost of these devices is approximately $2,000 to $4,000 
per unit. While they may not reduce present primary clerical costs, 
they do provide the data in a form which can be used by the com- 
puter without further processing clerical cost. 

In addition to these devices there are a number of others which 
have been investigated, some of which show considerable promise. 
Among these are point-of-sale recorders such as cash-register attach- 

tails on how combinations of "0" and "1" can be coded to represent numbers 
and the alphabet. A similar "two state system" is the Morse code. All numbers 
and the alphabet can be represented by combinations of dots and dashes. 



36 Electronic Computers and Management Control 

ments which will automatically prepare paper tapes as the register 
is used. Portable paper-tape punches are being employed to prepare 
input of inventory information at store and warehouse locations. 
The teletypewriter is being used to transmit input data over distances; 
the paper tape which comes off the receiving station can be read by 
a computer. In the department-store field, prepunched tags have 
been developed which can be used to transfer data to tapes or cards. 
In a document-sorting application a holder has been developed which 
can be read and sorted electronically while other information con- 
cerning the document is accumulated on tapes. 

Special devices are available which convert rotation speeds, tem- 
peratures, and similar data into digital form which can be recorded 
on tapes or read directly into the computer. Other devices which 
can read directly into the computer include such items as keyboards, 
turnstile and other counters, and dial telephones. A class of devices, 
already much in use for scientific measurements but not as yet adapted 
to business-data processing, comes from the area of telemetering. 
These devices transmit data over distance using high-frequency radio 
signals. 

Input media can be used for storing information. Magnetic tape 
can store more data within a given space than either punched cards 
or paper tape. For example, 1,500 feet of seven-channel tape can hold 
the equivalent of over 45,000 punched cards. The cards would require 
several file drawers. A seven-channel tape reel is less than a foot 
in diameter and less than an inch thick. 

After the data have been originally recorded on tape or cards, the 
media must be placed in a reading unit. Magnetic-tape-handling de- 
vices can read (and write) information at high speed— some at over 
15,000 characters a second— if the data are sorted in an order that the 
computer can use. If not, some tape-handling units can search for 
a particular item on the tape. This means the tape unit must read 
in both directions; interlocks are provided so that the unit will not 
try to read until it has found the address of the item wanted. Tape 
searching is usually slow; if items are widely separated on long tapes, 
it may take over a minute per item. 

Paper- tape input readers are of two kinds: those using mechanical 
brushes and those using photoelectric cells. The former are similar 
to teletype readers and operate at the same speeds— about 10 char- 
acters a second. The latter have operated at speeds of over 600 



Survey of Electronic Methods of Data Processing 37 

characters a second. They are now being offered for a few thousand 
dollars a unit. 

Internal Input — Storage. Not all the data on which the machine 
operates need be fed into the system from the outside. A great deal 
of information may be stored in the computer, so as to have it avail- 
able on very short notice— a small fraction of a second. Other data 
which are only used occasionally can be stored separately and placed 
on the reading device only when they are needed. An example would 
be payroll information kept on a set of magnetic tapes which could 
be removed from the computer except when the payroll is run. 

A program of instructions for the computer can be handled in 
either of these two ways: It may be kept in the machine if it is to 
be called on shortly, or it may be stored externally if it is not to be 
used in the near future. 

STORAGE DEVICES 

Storage, or the "memory" of computers, is one of the most interest- 
ing of the abilities of electronic data-processing systems. Funda- 
mentally, a memory system is a method for delaying information until 
it is desired at a later point in time. To accomplish this requires a 
physical system which can hold a state indefinitely, subject to rapid 
read-out, or which can delay the transmission of a signal. 

The delay period can be brief. It need be only a fraction of a 
second if the signal can be reconstituted at the end of the delay 
period and fed back to start a new cycle. This can be repeated many 
times, thus achieving the extended period of memory frequently re- 
quired. (It is believed by some that the human brain cell operates 
in this fashion.) 

Paper tapes and punched cards are efficient memories in the sense 
that they are a physical state which can retain symbolic data for long 
periods with no need for reshaping and strengthening the record. 
However, the read-out is comparatively slow and the information 
stored cannot be changed in the memory. 

There are a surprisingly large number of ways to store information 
magnetically or electronically so that it can be read out rapidly and 
so that the information can be changed if desired. There also are 
other compact storage media, such as photographic film, where the 
data can be read rapidly but which cannot be erased and reused. 



38 Electronic Computers and Management Control 

Most machines have "registers" which store the data temporarily 
as they pass through the computation centers. These registers are 
in the form of vacuum tubes, diode combinations, condensers, magnetic 
cores, or other devices. However, registers are limited in capacity 
and are not usually considered as "storage." Rather they are used 
as part of processing, or as "buffers"— delay points to accumulate data 
at input and output. 

Magnetic Cores. One of the best of the high-speed memory media 
is the magnetic core, which is a small doughnut of magnetizable ma- 
terial. The cores are wired together. The result looks something like 
a wire window screen with a tiny metallic ring looped wherever the 
wires cross, except that the wires are not permitted to touch each 
other. 

The ring-shaped metallic doughnut can maintain magnetism in 
either direction (see Figure 2). It has an interesting magnetic char- 





FlGURE 2 

acteristic (technically called a "hysteresis loop") in that the core re- 
sists attempts to change the direction of magnetism, up to a point. 
Beyond that point, when the core is exposed to a force trying to change 
its direction of magnetism— a force greater than an ascertainable 
amount— the core will suddenly reverse its field. It will then resist an 
attempt to change it back, until subjected to sufficient force. 

A set of cores can be wired in a matrix (only part shown of many 

wires and many cores) (see Figure 3). 
If a charge of a certain size is sent over 
wire 1, none of the cores will change 
direction of magnetism if the charge is 



ih O 



£p $ 2 below the amount they can "resist." If a 

charge is simultaneously sent over wires 
1 and 3, then only core A will reverse 
field, since it is the only core to receive 
a double "dose." The reversal can be detected by the circuitry. This 
setup can be used to store a binary bit of information in each core 



4 
Figure 3 



Survey of Electronic Methods of Data Processing 39 

and read it out when desired, by means of the instruction "address," 
which tells the computer which core or cores to locate. 

Core memories holding the equivalent of over 20,000 alpha-numeric 
characters have been built, in which the random-access time is 3 to 15 
millionths of a second. All the manufacturers of large general-purpose 
computers are using or contemplating the use of core memory. 

Other High-speed Storage Devices. Another form of rapid-access 
memory is the acoustic delay line. A pulse is fed into one end of a 
column of mercury. It travels to the other end in a fraction of a second 
longer than it takes the pulse to travel the same distance on a wire. 
When the pulse reaches the end of the mercury column it can be fed 
back to the beginning on a wire, and the cycle thus repeated, or the 
pulse can be read out for use in the computer, or both. 

The acoustic delay line makes the data available very rapidly— 
25,000 a second in the serial order that they come off the line, and 
at the slowest (if the machine must wait for an item to run through 
the column, or "tank") more than 2,500 a second. Each position of 
the items in the "line" of pulses in the tank is given an address, so they 
can be reached by the computer "instruction." 

Like other high-speed memory, capacity is limited for economic 
reasons. One large machine can store 12,000 characters. 

The line loses the data if the power is cut for a moment. However, 
if the memory is called upon to store data that are of lasting im- 
portance, or if a process lasts for some time, periodically a tape 
read-off is made of the information on the tubes or lines. Then if 
something causes a loss of the information on the high-speed storage 
mechanisms, the computation can be started up again from the point 
at which the tape read-off was made. 

Another form of storage which makes the data available with 
great rapidity, but which also loses the record if the power fails, 
is electrostatic storage. In this system a matrix of charges is planted 
on the face of the television-type tube by a moving beam, analogous 
to that which creates a television image. Any charge can be read by 
the beam action, or can be rejuvenated. This system can make a 
"word" of information (several dozen bits) available in 3 to 8 mil- 
lionths of a second. The capacity of the electrostatic storage in one 
computer, using 72 tubes, is about 20,000 decimal digits. 

There are many other forms of storage, including barrier-grid 



40 Electronic Computers and Management Control 

systems, holding beams, ceramic-coated materials, and potential 
systems. Many have extremely high speed— a millionth of a second 
being sufficient for the actual reading operation. Some highly ex- 
perimental systems have stored over a billion digits in the equivalent 
of milk-bottle size. Work on these systems usually is top secret, 
and commercial applications are hard to predict. 

Photographic film or paper shows promise for rapid-access memory 
where there is no need to erase the information. (This is possible in 
processes previously described.) If the data change, the variations 
must be stored in auxiliary storage until the main storage is re-created. 

As is known from the use of photographic film for astronomy, the 
area affected by a light beam can be remarkably small. These de- 
vices, are very compact. Methods of reading this type of storage 
are now under development. They show promise of enabling the 
storage of enormous amounts of information in a very small space; 
one commercial system can store data in the ratio of 90,000 to 1 as 
compared with printed-paper storage. 

The method has already been used by the Bureau of Standards 
as a means of sorting items. The code number of each item is stored 
on film, which is scanned by a photoelectric reader. 

All these systems offer at least moderate-speed "random access." 

Significance of Random Access to Storage. "Random access" is 
a term much used in the electronic field. An easily understandable sys- 
tem of random access now in operation is the telephone network. 
When you pick up your receiver, the phone company has no knowledge 
of the number you are going to dial. As far as they are concerned, it 
is a random choice, except that you are more likely to call a number 
nearby; so you are more closely connected to nearby phones through 
the switchboards and operators. 

An ideal memory would have immediate access to any data stored 
in it. The human brain is the nearest approach to this ideal. (The 
authors', however, are not good examples!) The larger the volume 
of data, the more it costs to make it rapidly available. Big computers 
have very fast access— in millionths of a second— to about 5,000 to 
20,000 characters. The cost per character, however, is at present 
several dollars, so this machine component alone may cost over 
$100,000. Using magnetic drums (see below), the computers can 
reach fifty to several hundred thousand items in hundredths or thou- 
sandths of a second. The cost per character will be well under a 



Survey of Electronic Methods of Data Processing 41 

dollar. The machines store very large volumes of data on tapes, 
where it may take several minutes to reach an item by running from 
one end of the tape to the other end, but the cost per character is 
almost insignificant. 

The storage cells with very fast access are used to hold such items 
as the program of instructions, which the computer calls on in 
rapid succession. The systems also use high-speed storage for 
temporary holding of intermediate stages in calculations. The rate 
at which these items can be reached (the program instructions and 
intermediate calculations) usually is the determining factor which 
sets the speed of operation of the computer. 

Cells with less rapid access, such as on the magnetic drums, are 
used to store instructions that are not being used at the moment but 
which may be wanted soon, as well as useful calculations— perhaps 
a table of functions or of withholding taxes. These may be trans- 
ferred into the rapid-access memory just before they are to be used. 

On the other hand, medium-size computers and special-purpose 
computers have only this moderately fast access to memory. It may 
be all they need. If they are dealing with human operators who 
cannot feed them information very rapidly, or if only simple calcula- 
tions are required, or if the logical design of the electronic system is 
geared to the needs of a particular data system, moderate storage 
speed may suffice. In such cases, very high-speed memories are a 
waste of money. 

Speed of random access adds considerable cost to the digital com- 
puter. If random-access speed is required for only a minor portion of 
the business operations, it may be best to use a moderate-speed com- 
puter plus a special system designed just to handle the specific high- 
speed requirements. 

Magnetic Drum or Disc Storage. Where the need is for storage 
of a fairly large volume of data, e.g., an inventory of many thousand 
items which must be available rapidly and in random order, one of 
the best devices now in use is the rotated surface of a large cylinder 
or a stack of discs. The surface of the "drum" or disc is coated with 
the same magnetic material as is used on tapes, and the reading and 
writing are accomplished with the same type of "heads." The heads 
may be movable or fixed. 

The drums and discs are rotated at speeds which vary somewhat, 
but which usually produce high surface speeds— over a thousand 



42 Electronic Computers and Management Control 

inches a second. Since the charges can be placed several hundred to 
the inch, as on tapes, the information is presented to the heads 
very rapidly once the surface has turned so as to bring the proper 
area under the reading head. This may take up to a fiftieth of a 
second for large drums but may be as low as one two-hundredth. 

Heads may be fixed, in which case they cover a given path around 
the drum, or may be movable so that they can be positioned over 
any of several paths. The latter system is slower, taking up to a 
second or so to move the head. Through proper placing of the 
heads it is possible to have several heads reading the same channel, 
thus speeding up the reading. This is useful on computers which lack 
high-speed memories. Systems which have a movable head covering 
several channels are generally not used for active programs but 
rather for volume storage of a filing-cabinet character. 

A method of increasing the volume which can be handled without 
greatly increasing the cost has been called "air-floating" the heads. 
In this type of system the heads are not fixed but can "ride" the sur- 
face of the disc or the drum on a thin film of air. Since this permits 
looser tolerances, a much larger surface can be used. The system 
has also been applied to wide tape "bands" which are, in effect, 
flexible drums, since they form closed loops, like a belt. 

As in all types of storage, choice of the area on the drum to be 
read is achieved by the instruction "address." In this case, the ad- 
dress specifies the channel and "cell number." A special circuit and 
reading head count the "cells" on a marked channel. The address 
enables the computer first to select the proper channel and then to 
match the count to the proper cell where the information is to be, 
or has been, stored. 

Magnetic drums do not lose their memory if the power is cut off. 
However, they may lose it if a circuit fails in such a way as to send 
a strong charge through the writing heads, or if a head works loose 
from its position a few thousandths of an inch away from the drum 
surface so that it gouges the surface underneath. ( The same casualty 
may befall a magnetic tape. ) Neither of the described events is likely 
to happen, but as a safeguard, whenever a particularly important bit 
of information is on the drum, it is also drawn off on a tape. 

Drums of moderate size— holding 10 to 30,000 decimal digits— have 
been less expensive than the high-speed memory units described 



Survey of Electronic Methods of Data Processing 43 

earlier, and due to their advantages of random access at fair speed 
to large amounts of information they have been incorporated into 
many computer systems. The exact size of drum required must be 
determined from the needs of the system. First consideration is the 
need for the type of service that they provide. This must be bal- 
anced with the needs for higher-speed units and for more sizable 
slower-access units such as the tapes. 

In several smaller and less expensive systems where high-speed 
access devices are not available, the drum is used to store the program. 
Drums can also be used to store some of the working data. 

There are experimental drums so large that they might be used for 
all storage requirements. More frequently the drum is used as an 
intermediate storage stage, with data permanently stored on tapes. 
Tape reels and tape trays are also used as storage. (See "Input.") 

Criteria for Selection of Storage. In selecting storage devices, the 
important criteria are the amount of information which can be stored, 
the speed with which it can be reached, and, of course, the cost of 
building and maintaining the system. Eckert, one of the pioneers in 
the field and a developer of the UNIVAC, has estimated that an 
increase of greater than 20 per cent in the cost for doubling the 
speed is usually unwise, but this opinion is not shared by all designers. 

Moreover, as indicated, it is necessary to have a balance in the 
speed. The arithmetic system should not have to wait for the memory, 
if it can be avoided economically. This is the province of the com- 
puter designer, since the businessman is primarily interested in the 
over-all speed and cost of the system. But it is well to remember that 
the system must have this balance— that extra payment for extra speed 
in a part of the system may be wasted money. 

Memory methods usually are identified by the type of medium 
that they employ for storage, since the various methods of using each 
type are fairly well known. Speeds for using a certain medium 
usually do not differ as much from computer to computer as do the 
speeds between various types of media. That is, two computers are 
likely to read their magnetic drums at speeds that are much more 
comparable than the speeds of either computer using drums com- 
pared with using punched cards or magnetic cores. 



44 Electronic Computers and Management Control 

OUTPUT 

Output can take two forms. One form is suitable only for further 
processing in the system. The other produces printed records or 
reports for human use. 

The first form is similar to input, using tapes or cards, since in 
effect each such output becomes an input if it is fed back into the 
system. The important media for this use have already been described 
under "Input," and this form of "external" storage will not be con- 
sidered further here. 

The other form of output is of special concern to businessmen 
because it is often considered to be one of the bottlenecks in present 
systems. 

Originally computers used electric typewriters for direct print- 
out. Some now use typical tabulating-card printers. Because the 
computer provides data so much more rapidly than the printers can 
operate, for some systems it has been thought advisable to plan for 
a battery of these printers. The buyer can employ the number 
which is economically justifiable in order to achieve the desired 
output. 

High-speed printers operating at 600 to 800 lines a minute and more 
have been shown, and several firms now offer such equipment. One 
of the difficulties holding back the use of these printers is the problem 
of moving and stopping the paper at high speed. The actual printing 
can be done very rapidly by a number of methods. Among there are: 

1. Mechanical printers (like the line-at-a-time printers now used 
in punched-card applications). 

2. Electrical printers using special papers that can be marked by 
electric charges (such as the "smoke" printer), or which record 
characters on photographic film at high rates of speed. 

3. Electronic printers that set up an image which must be photo- 
graphed (like the Charactron, which sets a "pageful" of in- 
formation on the face of a cathode-ray tube, from which it is 
photographed). 

The latter two types, when perfected, perhaps will offer the 
highest speed of operation, over 20,000 characters a second. Some 
of them may prove to be the best for special applications where very 



Survey of Electronic Methods of Data Processing 45 

high speeds are required. However, they are expensive, and as 
yet have not seen sufficient use for proper evaluation of their abilities. 

At present the mechanical printers appear to be most reliable 
and most economical. Some of these print with hammers striking 
rotating wheels, others with a matrix of wires which form characters 
by selecting dots from an array. Some of these print a line at a 
time, others a character at a time. 

In many cases the demand for high-speed output is partly derived 
from a reluctance to trust tapes and drums. As these media become 
more and more accepted, high-speed printing will be necessary only 
for such items as payroll checks, customer bills, magazine addressings, 
invoice preparation, and so on. For these the computer can be in- 
structed to prepare the same type of material that is now achieved 
with punched cards or Addressograph plates. 

Emphasis on speed is frequently encouraged by the electronic en- 
gineer, who thinks in terms of printing out everything processed as 
rapidly as the processor provides the information. This is spoken of 
as a "balanced" operation between components of the computer. 

However, a great deal of this print-out may be unnecessary. Re- 
ported data can be reduced through automatic analyses that are pro- 
grammed in the computer, so high speeds of print-out will not always 
be required. For example, a material-control system may only 
need data on those items which were not delivered by the vendor. 
The computer can provide this through automatic comparison of 
actual receipts with promised delivery dates. The details will still 
be stored in the media (e.g., tape) which can be printed out in 
detail if necessary. 

PROCESSING 

Control. All large computers have an operator's panel which is 
used to govern the operation of the machines. The control panel 
usually has buttons, keys, switches, and signal lights. Using these, 
the operator can start and stop the machine, give instructions as to 
which routine to start, alter instructions in a stored program, and 
"communicate" with the machine. By "communicate" is meant that 
the operator can read the pattern of signal lights in order to tell 
what is happening in the machine, can stop it, add instructions if 
desired, and can see the effect of these instructions as portrayed 
through signal lights. 



46 Electronic Computers and Management Control 

Since the control panel must be handled only by trained per- 
sonnel, we shall not attempt to describe it in more detail. The im- 
portant point is that a computer operation is not irretrievably fixed 
even after a program has been made up and stored in the machine. 
It can be stopped and the instructions changed at any time through 
the control panel. This is a desirable feature in any system which 
has to deal with occasional changes, a type of problem which cannot 
be handled with equal efficiency on many punched-card systems. 

When a system has to be changed, however, a new program usually 
should be prepared and stored in the computer, for the speed of the 
operation is greatly hampered by human intervention. The control 
panel is primarily used for maintenance and checking. 

Arithmetic and Logic. The logical design of the computer deter- 
mines the way in which the various components are linked together, 
and the way in which the machine adds, subtracts, compares, and so 
on. The design of the computer has a marked influence on the cost, 
reliability, and capacity of the machine. Appendixes 1 and 2 describe 
the logic of a computer which uses binary arithmetic. Other arith- 
metic and logic are available and have been used. The design also 
determines the number and variety of program instructions. Some 
large machines have over 70 types of instructions, some scientific 
computers fewer than 20. 

It is possible that the impact of improved logical design eventually 
may be greater than that of improved components. An example of the 
progress that has been made can be taken from the work of Floyd 
G. Steele. In 1948 Mr. Steele designed the Maddida digital differ- 
ential analyzer, a scientific computer. The computer had 100 tubes 
and 1,000 diodes. Through improvements he was able to develop a 
new design which reduced the parts required about 75 per cent. The 
new design, known as the Litton-20, is much more reliable in op- 
eration and requires much less power. 

While changes in designs for processing data do not reach the 
commercial market immediately, within the next five years the develop- 
ment of new components, linked by improved logical designs, will 
make computers and special electronic systems considerably cheaper 
and more reliable. 

The arithmetic and logical circuits of different computers vary con- 
siderably, but all of them operate at high rates of speed. The over-all 
speed of most computers is not governed by these parts of the system 



Survey of Electronic Methods of Data Processing 47 

to any great extent, at least not to a degree that has much significance 
for business purposes. Speeds of operation are related more to rates 
of input, access to storage, and output. 

In general, the user can plan to employ present electronic systems 
for all types of arithmetic and comparison usually found in an ac- 
counting system— to calculate payroll, bonuses, etc., to prepare vari- 
ances, summaries, and so on. Many machines, and especially the 
general-purpose machines, can perform mathematical manipulation 
of far greater complexity than is ever attempted in regular business- 
data handling. The machines can solve differential equations, invert 
matrices— in a word, they open up new horizons for business analysis. 

ELECTRONIC SYSTEMS AVAILABLE 

There are a number of electronic systems which can be obtained 
for business use. Most of these are complete systems, produced by a 
single manufacturer. Others are made up of components from sev- 
eral manufacturers, although one manufacturer usually produces the 
major part of the system. 

The specifications noted in Appendix 3 are taken from information 
which manufacturers have released concerning their own products. 
Of course, new developments appear frequently, so this material is 
not in any sense a catalogue. Rather, Appendix 3 should be viewed 
as another method of presenting what electronic systems can do— a 
further detailing of the material presented here. 

The order of presentation in Appendix 3 has no significance. Each 
of the manufacturers offers equipment with abilities which make it 
more suitable than the others for some purposes. No one system can 
be best for the whole field of uses. 

In evaluating the various electronic systems, the following con- 
siderations should be taken into account. 

A. Speed and capacity of system, at the following points: 

1. Recording. 

2. Input to computer. 

3. Computation and control. 

4. Access to storage ( volume of storage also very important ) : 

a. Random. 

b. Serial. 



48 Electronic Computers and Management Control 

5. Output: 

a. In storage forms for later input. 

b. Printed material. 

6. Sorting, collating. 

B. Flexibility of system: 

1. Ability to change its own instructions, and/or to operate from 
a series of stored instructions, as against use of a limited 
program built into the system and controlled by a keyboard. 

2. Variety of instructions offered. 

C. Cost: 

1. Cost in terms of amount of data processed by each component 
in a given period, as well as over-all cost. 

2. Operating costs, which include machine rental or amortiza- 
tion, personnel, space requirements, air-conditioning, main- 
tenance, power, supplies. In most cases cost estimates can be 
obtained directly from the manufacturer. 

Costs mentioned in Appendix 3 are subject to change and should 
only be interpreted as indicative of the size of the system. 

Almost all of the equipment described is available under leasing ar- 
rangements. 

Considerations other than those listed may be important in indi- 
vidual circumstances. Accuracy and dependability are significant, of 
course, but except for experimental machines, commercial equipment 
now rates very high in almost all instances. Since there is a constant 
effort to improve the machines, the degree of reliability may depend 
on the current state of development. 



CHAPTER FOUR 

Studies and Applications of Electronic Systems 



Electronic computers applied to business-data processing are a 
development of the years since the Korean War. Deliveries of large- 
scale and medium-sized electronic systems in quantity began in late 
1955. Estimates indicate that demand may exceed supply for some 
items at least until 1960. However, there have been sufficient re- 
search studies and actual applications so that businessmen now are 
able to learn a great deal about the selection, installation, and op- 
eration of electronic systems. 

Applications of electronic computers to business-data problems 
already can be found in a number of functional fields such as account- 
ing, production scheduling and control, and inventory control. Ap- 
plications to special problems exist in particular industries. Hundreds 
of companies are actively studying the possibility of using the new 
electronic systems. Hundreds more have ordered electronic systems 
or have given "letters of intent." 

Nevertheless, because of the newness of the electronic tools and 
because of the size of financial risks involved, a number of com- 
panies are still waiting until others have pioneered in this field. One 
businessman has publicly described his position with the old saying: 

Be not the first by whom the new is tried, 
Nor yet the last to cast the old aside. 

This cautious attitude, frequently the wisest course of action for 
businessmen when confronted by a new field, may not be advisable in 
the area of electronics. Since it requires a lengthy period of study 
and analysis to prepare for the computer, those firms that wait too 
long for others to pioneer may well find themselves in a difficult 
competitive position. 

49 



50 Electronic Computers and Management Control 

Those who undertake this type of research find, besides other 
benefits from using new machines, that cost savings may be im- 
mediate. Men are stimulated to seek better systems and procedures, 
an outcome of the challenge presented by the result of their studies. 

Enthusiasm can be overdone; some may even be experimenting 
with electronic devices for the pure joy of working with a new gadget. 
But more often the computer is regarded primarily as a useful tool, 
one which may help a company to face changing markets and 
stiffening competition. 

GENERAL CHARACTERISTICS OF ELECTRONIC 
DATA-PROCESSING APPLICATIONS 

Electronic Data-processing Systems Usually Are Selected After 
Feasibility Studies. Orders for computers usually have been placed 
only after extensive preliminary studies. The primary purpose of 
these studies has been to determine whether the electronic equip- 
ment was economically feasible. The studies showed that manage- 
ment not only would receive more timely data, but also that they 
would be able to release clerical help for other duties. In addition, 
the company would be prepared to handle more information as the 
volume of transactions increased in the future. 

Charles E. Becker, president of the Franklin Life Insurance Com- 
pany, has stated, with reference to the company's order of a UNI VAC: 

Our studies revealed that this electronic system will speed many of our 
administrative functions to an extent which at first is difficult to grasp, while 
at the same time releasing valuable clerical help for other duties. 

We are convinced that the insurance field in general will not be able to 
keep up with the continued demand for its services without introducing 
electronic data processing equipment to handle the heavy volume of paper- 
work which future expansion will bring. The physical problem of finding 
enough room for the employees and office equipment needed would be 
staggering in itself. 

We have been in process of converting our procedures to electronic 
requirements for some time now. Four primary functions will go on the 
UNI VAC first: premium billing and accounting, valuation, agents' com- 
mission calculating and accounting, and dividend accounting. Others will 
be added as we go along. 

In premium billing and accounting the UNIVAC will maintain the 
master file, select notices due and prepare them so that they will only have 



Studies and Applications of Electronic Systems 51 

to be placed in a window envelope for mailing, and handle the payments 
as they come in. In valuation the system will take care of the computation 
of reserve liability. All accounting necessary in the other functions will also 
be handled automatically. 

Current plans for all four of these functions, based on present volume, 
will be accomplished in daily operations totaling 10 to 12 hours, five days 
a week— one-and-a-half shifts. The UNI VAC staff will total 20 people for 
both shifts. At present, this work requires 1200 clerical hours of work 
every day. 

A special team of Franklin Life personnel is being trained to handle 
the UNI VAC system when it is installed in our offices. In a sense we feel 
like pioneers, for while this so-called electronic brain has proven itself in 
other applications there is no history for us to go by in our field. 

We have no qualms about this, however. On the contrary, we are 
anxious to get the "electronic revolution" under way in our offices, for we 
are certain our future would be restricted without it. 

In one of the first studies made, Prudential Insurance developed the 
following estimates of how it can cut use of office machines : 



Numbers and types of machines 

used by headquarters to run its 
ordinary policy service 
21 Key punches 
12 Verifiers 
31 Sorters 
21 Collators 
16 Reproducers 
11 Interpreters 
16 Accounting machines 

7 Summary punches 

2 Calculators 

4 Miscellaneous machines 



Numbers and types of machines 
used with the IBM 702 

21 Key punches 
12 Verifiers 
11 Interpreters 
1 IBM Type 702 
electronic data- 
processing machine 
(containing about 
25 smaller machines) 
7 Summary punches 
4 Miscellaneous machines 



The new system must service around 3 million accounts receivable. 
In so doing, each year it will send out 10 million premium notices and 
account for 10 million remittances. About 80,000 policies a day will 
receive servicing. 

The system has been refined further since these early estimates. 
Prudential is locating an IBM 705 in each of several regional head- 
quarters. 



52 Electronic Computers and Management Control 

E. F. Cooley, associate director of methods research of the Pruden- 
tial Insurance Company, has declared: "There's no shortcut in deter- 
mining whether a big tape-operated system is more economical than 
a punched-card system." Intensive study beforehand is necessary, and 
even then, actual operation may alter the findings. 

Gradual Conversion to Electronics Is Feasible. An example of a 
company which adopted the policy of gradual installation is the 
Metropolitan Life Insurance Company. The company examined 
a number of functions before it found one in which the use of an 
electronic computer was clearly justified. When this was found, 
however, and the application was coupled with uses in the other 
functions examined, the possibilities of an electronic computer be- 
came so attractive that the company now contemplates wide use 
of the UNIVAC. It is investigating the advantages of several in- 
stallations. 

The function which at first seemed like an obvious application was 
in actuarial investigations. However, study showed that, while the 
computer could do the work, it would not pay to have the machine 
for this purpose alone. The actuarial computations involved were 
not made frequently enough to justify the cost of an electronic system. 

Another field was policy settlements, such as death claims, cash 
surrenders, and changes from one plan to another. In policy settle- 
ments there is need for frequent reference to tables of unit cash 
values, tables of single premiums, etc. While it was possible to pro- 
gram a computer to calculate the proper values, the volume of this 
work alone was too small to support a computer, especially since the 
computer would require storage capacity that was quite large com- 
pared to the use value of the information. 

File keeping on magnetic tapes was also examined. Policy informa- 
tion could be kept on tapes, but on investigation it was found that no 
machine existed with sufficient random access for this type of storage. 

The area which provided the largest clerical savings was regular 
policy service. The policyholder must be notified when the premium 
is due, the amount of premium, and the amount of his dividend. The 
computations involved, while simple, are in large volume. The sys- 
tem must provide regular notices and statements to the policyholder. 
However, it became clear that in order to use an electronic computer 
effectively in this area, very extensive changes would be necessary. 
The company wanted to acquire firsthand operating experience before 



Studies and Applications of Electronic Systems 53 

any far-reaching changes were made; therefore it looked elsewhere. 

It found the area of work it was seeking in its actuarial division. 
A large amount of routine repetitive work is done there to develop 
the actuarial statistics needed for the company's financial statements 
and for various analyses of its experience. Investigation disclosed 
that it could gainfully apply an electronic computer to this work, 
so the company arranged for the UNIVAC to do the job. Thus, the 
task of producing actuarial statistics, which involves large volume but 
little computation, was found to be more productive than actuarial 
investigations and its related computing requirements. 

The company regards this area as a particularly good one. It was 
localized to one division. It involved a minimum of changes before 
the company could get started. No policyholder service would be 
impaired if the operation failed. Since the work for which the 
electronic computer substituted was already highly mechanized, it 
required the computer to establish its superiority over punched-card 
techniques rather than over clerical procedures— a much more difficult 
thing for it to do. The annual rental of machines initially replaced 
was over $200,000. 

Because Metropolitan Life operates on a highly centralized basis, 
with much of its work concentrated in one location, the large volume 
of work necessary to justify an electronic computer was readily 
available. Smaller companies, and those whose activities are more 
widely separated, may not be as fortunately situated. 

Metropolitan Life chose an application for first installation where 
difficulties or delays incurred in the process of installation or early 
trial runs did not affect the normal operation of the business. The 
company was able to train its employees in computer technique. 
Actuarial work requires the use of capable actuaries, who made the 
transition to computers more easily than nontechnically trained peo- 
ple. Once experience was gained on this application, others of a 
nature more concerned with temporal deadlines and with personal 
relationships were introduced. 

The experiences of the Metropolitan Life Insurance Company have 
been repeated in other companies. They have found that the elec- 
tronic computer can justify itself economically on simple repetitive 
routines that are not complicated but voluminous. 

There exist certain problems in insurance applications which re- 
quire special consideration and which illustrate the necessity for a 



54 Electronic Computers and Management Control 

careful study of all the side effects of an adoption of electronic proc- 
essing systems. For example, there is some question as to whether 
information recorded electronically will be acceptable to the various 
state courts and insurance departments. Only recently have some 
states accepted printed copies of the annual financial statement for 
insurance companies. Some courts still will not accept microfilm 
records. The company may need a visual record of the policyholder's 
account, which the computer will have to print out. 

For other applications, such as filing, company officials have ex- 
pressed some worry about the fact that a machine cannot exercise 
judgment if there is an error. For example, if a number on the policy 
is slightly incorrect, or a John Smith is recorded as John Smythe, the 
machine cannot look up plausible alternatives and find the data as 
easily at present as can a human clerk. 

More than One Function Will Be Handled by the System. Almost 
all companies that have ordered large-scale general-purpose com- 
puters expect that eventually they will put more than one function on 
their machines. 

George Troost, vice-president of Chrysler, has aptly summed up this 
attitude: 

We believe it is necessary to develop an integrated approach wherein 
the requirements of sales, procurement planning, material control, produc- 
tion, statistical, accounting and Treasury functions are viewed together 
so that a consolidated and coordinated system of data processing can be 
thoroughly developed. 

The concept of integration lies at the heart of one of the most 
comprehensive programs undertaken by any company in this field to 
date— the "common-language" project of United States Steel. De- 
scription of various aspects of this system is given in other parts of 
this book. It has also appeared in a variety of other publications, 
such as those of the American Management Association. 

Another company which has a comprehensive program is Sylvania 
Electric. Early in 1954 Sylvania established a team under the super- 
vision of the controller. Its assignment was to make an evaluation of 
every piece of equipment available, or likely to be, and of every in- 
stallation which could be visited. 

The basic philosophy adopted was that the company should estab- 
lish a data-processing and communication center. The location was 



Studies and Applications of Electronic Systems 55 

to be chosen solely on the basis of the communication needs of the 
company, the available communication systems of A.T.&T. and West- 
ern Union, and desirability as a place where the company could get 
the necessary personnel to be willing to work. It was not to be in- 
fluenced by any particular production location, but to serve several 
dozen factories in all sections of the country (with the possible ex- 
ception of the West Coast). 

The communication center was to serve the needs of decentralized 
management, not to centralize the authority-responsibility relation- 
ships of the company. These needs were believed to be character- 
ized by one dominant factor: In recent years competitive conditions 
have compelled management to continually shorten the time period 
between events and the receipt of information concerning the events. 

The company installed a UNIVAC in 1955. It was estimated that 
it might take 2 years at best for the installation to pay for itself, but 
in the last analysis this was not the vital factor. They were aiming 
to get next-morning reports on: 

1. Exact inventory, by location. 

2. All sales. 

3. Production. 

If the machine could aid in stopping "dog runs"— production of 
unneeded items— then the saving would be far greater than any an- 
ticipated from clerical functions. 

Don Mitchell, president of Sylvania, has declared that the advent 
of electronics will "separate the men from the boys" in management. 
In the past, too often a manager could put off a decision with the 
excuse that he had to have time to gather more information. Now, 
everyone will know that he can gather the information rapidly, and 
it will be up to him to make decisions when they are called for. 

Pacific Mutual, one of the first to install a computer, came to the 
conclusion that they could not justify the use of a large-scale machine 
unless they placed more than one function on it. Some of the 
functions would have to accept changed routines. For example, to 
answer certain types of inquiries it now would be necessary to wait 
until the next morning, since the material would be stored on tapes. 
It is not economical to run some of these more than once a day. 

A related aspect of the Pacific Mutual installation is of consid- 
erable interest. At the time the company installed its UNIVAC, 



56 Electronic Computers and Management Control 

in the fall of 1955, the new UNIVAC II, featuring use of magnetic- 
core memory, had become available. Nevertheless the company de- 
cided to continue with the first UNIVAC, at least for the time being. 

The first UNIVAC uses acoustic delay lines for high-speed memory. 
Magnetic cores are faster and somewhat more reliable. However, 
they cost more. Pacific Mutual decided that though the acoustic 
delay line was technologically obsolete, it was not economically in- 
ferior. It would provide all the speed of computation which they 
required. If more high-speed storage should become necessary as 
use of the computer increased, the memory could be converted. 

Incidentally, Pacific Mutual hired an expert for full-time assignment 
to the task of answering the numerous inquiries about their installa- 
tion. 

Transition to Electronics Is a Time-consuming Process. Detailed 
installation studies often require 12 to 18 months for one application. 
The first application generally takes the longest. Personnel must be 
retrained to program a computer or else new people skilled in pro- 
gramming must be hired and trained according to the company's 
requirements. 

General Electric's application study for payroll at the Louisville 
plant began in September, 1953, and the first electronic payroll was 
run late in 1954. Six months later substantial improvements in the 
program were still being installed. 

Four procedures analysts and two outside consultants made up 
the first group. The work was divided into the following categories: 

1. Day work and individual incentive. 

2. Group incentive. 

3. Computation of gross pay, premiums, and other items, for both 
hourly and salaried workers. Distribution of labor cost and 
maintenance of employee files were included. 

4. Computation of net pay, including all taxes and other deductions. 

5. A general study, which covered the reports that were required, 
desirable changes in number codes, etc. 

The initial group was expanded several times. The exact size of 
the working force is difficult to assess, since the project was used to 
some extent to train analysts and programmers for the company and 
for outside consultants. 

Since major responsibility remained with operating divisions, it 



Studies and Applications of Electronic Systems 57 

was necessary to form committees to keep them informed and to give 
them a voice in the design of the system. This gave the computer 
group a chance to educate the other men in the workings of an 
electronic system and helped reduce opposition to the change. 

Controls were developed at the point where documents originated. 
The purpose was to ensure that all documents were received in the 
data-processing center, and also that all information received was 
processed by the computer. Documents were batched, given a con- 
trol number, and the time and identification were logged. 

After the materials went through key punch (which was necessary 
for input), the cards were balanced against a control card on an 
accounting machine. The control clerk then checked the total with 
the payroll section to make sure all the data forwarded were received. 

Cards were then converted to tape. A tape-control total was 
prepared and checked against the card control. The first computer 
run was to sort the data; during this run totals were prepared to 
balance the data on the tape against the control total. 

The extent to which such controls are required cannot be estab- 
lished without study of a particular operation. Because of the nature 
of an electronic system, introduction of faulty data is often quite 
costly, so that fairly elaborate control of input is often justified. Es- 
tablishment of such systems takes time. 

Electronic Systems Require Detailed Descriptions of Methods and 
Procedures. Because computers are not "brains," they must be given 
detailed instructions in order to carry out their function. Companies 
that already have good methods and procedures will have some 
advantage. However, very few companies have found that their 
systems and procedures are sufficiendy detailed for electronic re- 
quirements. Illustrations of this will be given later in this chapter. 

General Conclusions concerning Installation Studies. The follow- 
ing general conclusions have been reached by a number of companies 
that have pioneered in electronic systems: 

1. Over-all predesign of the program should be as complete as 
possible before the detail-flow charting and coding are under- 
taken. Changes tend to "cascade," like dominoes. 

2. A computer can handle all sorts of exceptions, but unless these 
recur with some frequency it is uneconomical to program and 
code for them. 



58 Electronic Computers and Management Control 

3. Sorting the information so as to report it in various sequences is 
usually an expensive process. (Development of better sorting 
abilities or larger internal storage may relieve this problem. ) 

4. Correction of errors should be designed to be handled as part of 
the system. Otherwise reruns, and sometimes reprogramming, 
may be required. The errors should be located as early as pos- 
sible to prevent compounding. 

5. Programs must be tested thoroughly before they can be used 
without difficulty. 

6. Achievement of pay-off in complicated accounting routines is 
not an easy task. 

7. Coordination between members of the study group must be 
of a very high order. Their efforts must be integrated. 

The remainder of the chapter is devoted to a more detailed descrip- 
tion of studies and applications in particular functional areas. 

APPLICATIONS IN ACCOUNTING 

Accounting applications include payroll, billing and accounts re- 
ceivable, disbursements and accounts payable, and cost accounting. 
However, in studying these applications, it always must be remem- 
bered that there is no hard and fast line between the different types, 
such as between payroll and certain phases of cost accounting. Nor 
is there a line between "accounting" applications and "inventory" 
applications. The distinctions are made here primarily to achieve a 
useful grouping of examples in the presentation. 

Payroll. Payroll computation has been a favorite choice for a num- 
ber of companies. The expense of present procedures, plus the fact 
that in many cases these are already mechanized, has led many ex- 
perts to believe that payroll should be one of the first to be placed 
on the computer. 

Experience, while limited, has shown that payroll can be handled 
electronically and that the systems can be made economical. How- 
ever, the many considerations involved in any fairly complicated pay- 
roll make it appear unlikely that substantial savings can be realized 
in this area soon after the installation is made. A careful study of the 
system is an absolute necessity, and this is both time-consuming and 
expensive. 



Studies and Applications of Electronic Systems 59 

Such firms as General Electric, U.S. Steel, and Ford have favored 
payroll as the first application for their actual or potential business- 
data processing system. There are several reasons why payroll may 
be chosen in spite of the risk indicated. Payroll, as a system, is more 
or less complete within itself. The volume of items is large. Current 
clerical costs are substantial. There are many computations, and 
great accuracy is required. There are elements of the data which must 
be stored, changed frequently, and reported out rapidly. Payroll is 
frequently handled now by IBM CPC's, or 604's, which are electronic 
computers of an intermediate character. 

Payroll data are used for a variety of reports, each report embodying 
somewhat different classifications of the same basic data, and each of 
which must be prepared regularly. This type of multiple-purpose 
processing is one of the electronic computer's best applications. 

Cost Studies. One study showed that a company with about 15,000 
employees and a weekly payroll spends approximately $545,000 a year 
for payroll and for processing of labor reports. Of this, about $240,000 
was for rental of punched-card equipment, almost as much again was 
spent for tabulating personnel, and another $75,000 was for divisional 
clerical costs to re-sort the tab reports for various management uses. 
It was estimated that an electronic computer could do the same work 
for a rental of about $250,000 a year, plus about $120,000 for labor 
costs, thus netting a saving of about $175,000 a year. 

The present payroll and labor distribution system takes about 10 
days, while only 3 days would be required by the computer. Since 
this would enable the computer to be used for other applications, 
the company estimated that they might save as much as another 
$500,000 a year. Even after discounting this somewhat, because of 
possible overenthusiasm, there is a margin of savings sufficient to 
warrant computer installation. 

A recent study undertaken by another large company disclosed that 
it could justify the purchase of a computer for its payroll alone. The 
computer could prepare the payroll, involving 140,000 to 150,000 
men, in about 3% hours per day (about 85 hours of computer time a 
month). In this connection it should be noted that most of the 
computers now in use are on a 24-hour-a-day schedule, with only a 
few hours down time for maintenance and repair. With sequence 
programming only a few people need be present to handle the actual 
operation of the machine on late shifts. 



60 Electronic Computers and Management Control 

In assessing these estimates, it also should be remembered that 
the experience of most installations has been that time approxima- 
tions were overly optimistic. This holds for both programming time 
and operating time. 

The above studies assumed continuation of the same methods of 
recording workers' time on timecards. Not much experimentation has 
been reported in this field, and there is some question as to when it 
will pay to replace this part of the system. Eventually, however, the 
timecard no doubt can be eliminated by using another form of input, 
such as paper tape or magnetic tape produced directly by the time 
clock. Or the information could be sent by direct wire from the clock 
to a central computer where the processing could take place imme- 
diately. 

Test Runs. Because of the importance of accuracy for payroll, it 
has been considered advisable to test-run the two systems in parallel— 
the old system in use, and the new electronic system. Such a proce- 
dure helps to get the defects out of the system before the switch-over 
is made. One study estimated that a complete change from first study 
to final change-over should take about 1 year. (Other studies have 
shown a year to be somewhat long.) Of this time, approximately 3 
months was to be spent on program testing and on making several 
experimental runs. In general, studies show that at least 6 months 
should be scheduled to program, code, and test the electronic payroll 
procedure. 

Because of this preparation, the cost of installing the computer is 
generally equal to the cost of the computer. As more experience is 
gained in this change-over, it seems reasonable to expect the time will 
be reduced, but the cost will continue to be substantial. 

There are a number of reasons why such a lengthy period of time 
is necessary. One difficulty arises from the fact that most systems in 
existence for payroll have been developed in line with various opera- 
tional functions and not as one integrated payroll system. As a result, 
before the payroll can be programmed on the computer, it is necessary 
to make a detailed system study and to integrate the various payroll 
procedures. This type of study requires the cooperation of engineers 
or programmers who know the attributes of the electronic machines, 
but who seldom know much accounting. Since most accountants know 
only parts of their payroll systems, and few of them have learned the 
attributes of a computer, such a systems study obviously takes time. 



Studies and Applications of Electronic Systems 61 

Once a study has been made, programming and coding the payroll 
system for a computer take additional time. The programmer should 
take into account many of the various exceptions which are handled 
manually at present, outside the mechanical system, since the ability 
to handle these is one of the advantages of a computer. A high level 
of competence is required to be able to work these exceptions into a 
system. The integration must be complete, for the computer must be 
told exactly what to do with any of the exceptions it handles. A testing 
period to correct the program and codes and to provide for any over- 
sights has proved to be necessary for all such complicated types of 
applications. 

Application Experiences. In the General Electric installation at 
Louisville it was originally estimated that a complicated incentive 
system for about 8,000 men could be handled by the computer in a 
few hours a week. This proved to be an underestimate. 

It required over 20 man-years to prepare and code the GE program. 
Even after several months of operation there still were difficulties with 
some items. The original program for the computer, including excep- 
tional payroll situations such as special deductions, required over 200,- 
000 instructions. To run such a program required about a full shift for 
a week. Some exceptional situations were determined to be inefficiently 
handled by the computer using the extended program, and were again 
given individual clerical handling. This change together with im- 
provements in the program has reduced running time considerably. 

Part of the computing problem at General Electric was due to the 
fact that there were three major methods by which hourly rated em- 
ployees were paid: day work, piecework, and group incentive. Any 
one man might conceivably work on all three during a given week. 

The General Electric approach was to devise a system by which the 
computer would be used to compute the payrolls and write the checks, 
leaving the responsibility for the remainder of the functions still with 
the various departments. This attitude was strongly influenced by 
the fact that the computer was to serve five decentralized departments. 
Therefore it was decided not to establish a central payroll section built 
around the computer. How much this influenced the success of the 
installation is difficult to determine. 

An important by-product of any intensive system study is the im- 
provement which can be achieved in basic methods and procedures. 
This may go a long way toward covering the expense caused by the 



62 Electronic Computers and Management Control 

necessity for lengthy study. Subsequent applications, as well as the 
necessary changes that will occur in the payroll system, eventually 
should be handled with greater speed and less expense as a result of 
the improvements that can be made in the computer itself and in the 
methods of staking out and attacking the problem. 

An example of the results achieved by an intensive system analysis 
is National Tube, of U.S. Steel. Although the company had spent a 
number of years in establishing the same basic labor-accounting sys- 
tem in each of its plants, various local modifications had been per- 
mitted. These were not significant as long as punched cards were used. 
When the electronic system was introduced, the payroll system had to 
be restandardized. Otherwise the programs for the computer would 
have been unwieldy, or reruns of the data would be required. 

An important result of such a study is the number of desirable items, 
not susceptible to mechanization on punched cards, which become 
simple by-products of properly designed computer runs. Examples: 

1. Determination of vacation and holiday eligibility. These are 
based upon average earnings during a reference period. 

2. Analyses of absenteeism, tardiness, overtime, etc. 

3. Reviews of stewardship. National Tube calls this "quota control." 

4. Automatic denomination of payrolls paid in cash. 

General Conclusions on Payroll. Once the data are available for 
machine use and suitable programs have been developed, pay checks 
can be printed, and payroll ledgers and reports that are prepared con- 
cerning payroll can be made up automatically. Changes in parts of 
the report requirements can be given the machine without changing 
the other parts of the program. 

In considering the desirability of using an electronic system for pay- 
roll, it should be remembered that a primary benefit of the computer 
is its use for related applications, such as labor-cost distribution, and 
labor reporting of statistics such as turnover, status of employees 
assigned to a given department versus the authorized number, and so 
on. Data for all of these can be accumulated simultaneously, thus in- 
creasing the speed of preparation considerably. Some initial studies 
made in a company with well over 10,000 workers and several hundred 
products showed that the computer could prepare detailed labor 
reports in somewhat over three hours. 



Studies and Applications of Electronic Systems 63 

Once the basic data have been recorded on a tape, the computer 
has considerable flexibility as to the type of report which it presents. 
Installation of a new report will require careful study, with a time 
period of as much as 1 to 3 months. But when the report programs are 
ready and available, the computer can gather the data for a number of 
reports simultaneously, and can switch from the preparation of one 
report to the preparation of another automatically and much more 
rapidly than is possible with card systems. 

Since payroll preparation must meet a fixed deadline, reliability is 
a major consideration. Computers do not have a long history of relia- 
bility, like that of punched-card machines. A computer can do many 
things, but it does place a lot of "eggs in one basket." Therefore most 
payroll studies include consideration of the desirability for the com- 
puter to have stand-by facilities in case of breakdown. This stand-by 
may be another computer some distance away, since the tapes, etc., 
can be flown great distances overnight, and the payroll checks can be 
similarly returned. 

Concern over breakdown has been minimized by the testing and 
modular developments which have been described previously. As for 
the speed of output, although high-speed printers are desirable, the 
same result can be achieved through use of several slower printers. 
In some cases the tapes can be converted to cards and a typical card 
printer employed. 

Receivables and Payables. There are numerous studies under way 
which are concerned with the application of electronic systems to 
receivables and payables. It has been found that, although study of 
other companies' experience can be informative, no application can 
be adopted by another firm without change. For each application 
there must be detailed investigation of the individual situation. The 
basic elements are fairly similar in many applications, but the specific 
types of information needed, the different points of origination of data, 
etc., must be established precisely. 

A study now under way at a leading oil company concerns the prob- 
lem of handling sales of gasoline on credit cards. The company proc- 
esses by hand several million slips of paper a month. A large amount 
of clerical work is involved in sorting the slips that come in at various 
times and from various stations. The company is experimenting with 
a method which would record the slip on film for the company's per- 
manent record. This would be followed by addition of metallic spots 



64 Electronic Computers and Management Control 

in a coded arrangement, automatic sorting, then billing to customers, 
and totaling sales for each station. 

Because this company's policy is to have the sales slip returned to 
the customer with the bill that is mailed to him, it is necessary that 
the original paper evidence be sorted also. Oil companies have found 
that a typical customer does not remember all his purchases and may 
doubt the total billed to him without the signed sales slips. The prob- 
lem, of course, has been to find a simple and inexpensive method of 
doing this. Several companies have experimented with procedures 
that mark the slips of paper so that they can be read and sorted, then 
processed for billing. 

A more direct system, known as Scandex, has been developed to 
read original evidences photoelectrically. A tabulating card, similar 
to those now used for this purpose, is prepared at the point of sale. 
The card is imprinted with data from the customer credit-card identi- 
fication. 

When delivered to the data-processing center, a photoelectric reader 
scans the information, verifies the account number, then automatically 
key-punches the scanned data into the original invoice. An adding ma- 
chine can key-punch the dollar amount of the sale. 

Estimates show that equipment renting for $1,000 a month (includ- 
ing service and maintenance), should show a gross saving of about 
$1,000 a month per million invoices. The system can be applied to 
situations where point-of-sale data can be controlled, such as with 
customer identification by credit plate as used by department stores. 

Applications— Utility Billing Procedures. The large volume of cus- 
tomer billing required by public utilities obviously offers a major 
opportunity for cost saving and improved handling in the utility in- 
dustry. A number of different studies have been made of the potential 
use of electronic systems in public-utility applications. First installa- 
tions are currently under way. 

Complete information concerning each customer's accounts can be 
kept in a main file. Each account shows the customer's name and 
address; the identification of all meters on the premises; the revenue 
class and the business code for statistical analyses; the last reading of 
each meter; consumption and demand figures for the last few months 
for each meter ( for commercial accounts the amount that the customer 
might demand is a factor in calculating the bill); the gross amount of 
the bill; the net amount; the due date and the amount of any bills 



Studies and Applications of Electronic Systems 65 

not paid; the current balance of the account; the description and 
amount of any miscellaneous charges or merchandise charges; the 
amount of the deposit, if any; and notations for credit actions, if any, 
taken in the past. 

In a computer system the accounts will be processed in order, by 
account number. The number, for example, may be a code developed 
from the reading day, the section of the city, route, block, premises, and 
individual customer. The tapes can be kept separate for each reading 
day, so that a partial sorting would thus be achieved. If the readings 
come in the same order as on the tape record, the sorting is facilitated. 

One pass through the tape input will be sufficient to put the new 
data from the meter readings into each account, while at the same time 
the consumption is calculated, a meter constant multiplier is applied, 
and the consumption is compared with past consumption to see if it is 
reasonable in the light of past experience (this would tend to catch 
reading and transcription errors). Where necessary, the readings for 
several meters can be added to determine consumption and adjustments 
can be made for periods greater or less than the scheduled billing 
period and for readings that indicate the consumption for more than 
one family unit. 

If a reading is missing, the machine can calculate and average for 
the immediate past period and make up a bill. Meter readings which 
appear to be out of line can be automatically segregated and listed 
for later investigation. 

As the changes and payments are entered into the main file, a pro- 
gram of selection can be watching for all items of unusual nature which 
should be treated immediately, such as bills past due or service dis- 
connect ordered but not performed. These data can be drawn off and 
printed separately. Records for accounts requiring off-schedule final 
billing are also drawn off. Payments recorded to the wrong account 
are noted in a separate list. Information necessary for the preparation 
of new meter-reading sheets is made up by the computer, which can 
print them out when desired. 

Receivables— Commonwealth Edison. One of the most advanced 
computer applications for billing is at Commonwealth Edison, in Chi- 
cago. W. E. Eggleston, machine procedures accountant, began his 
investigation of electronic possibilities in 1949, before any electronic 
machines were available or even certain to become so. 

Commonwealth Edison serves 1,800,000 electric utility customers in 



66 Electronic Computers and Management Control 

the Chicago area. They have scheduled the meter reading and billing 
on a bimonthly basis. The typical monthly bill for a residential town 
is under six dollars, so the billing and record keeping per dollar of 
gross income are important cost factors. 

The company had developed a highly mechanized system for cus- 
tomer billing. Meter readings were recorded on mark-sense cards. 
Bills were computed by IBM 604's. The meter card became the re- 
ceivable ledger record. The returned stub, which was prepunched, 
was processed mechanically, including the reduction of the receivable 
ledger. Any remaining ledger cards were added to the next meter 
card to produce the new bill. In this punched-card system only two 
points required clerical handling in detail— meter reading and accept- 
ance of payment. 

After careful studies it appeared doubtful that a large electronic com- 
puter system could compete with an efficient punched-card system if 
it was restricted to the functions just described. It was necessary 
therefore to enlarge the computer application to include the "excep- 
tions." 

In a punched-card system, it is necessary to maintain a large clerical 
force to handle these "exceptions": separate files for advance deposits, 
separate records of collection on slow-moving accounts, and adverse 
credit information. Inquiries about accounts by customers who have 
lost their bills are so frequent as to be a major factor in building the 
accounts-receivable system. 

The effort to build an economic system to handle these "exceptional" 
items has taken several years. At one point it was estimated that to 
print out all the information that conceivably might be called for each 
day would require an acre of printers, working double shifts. Or to 
have all the information immediately available on drums instead of 
tapes would call for a yearly rental of some $300,000,000! 

Preparation for the electronic system made it necessary for Common- 
wealth Edison to reexamine its detailed procedure. The solution 
adopted was to use tape records instead of cards for customer data. 
The tapes contain the name and address, metering and rate informa- 
tion, details of money due, and credit and collection standing. Basic 
customer tapes are reproduced three times a month on a cycle basis 
to spread the work load. The new tapes are used to print out an open- 
balance list which omits much of the fixed data but is sufficient for 
most of the required account reference. 



Studies and Applications of Electronic Systems 67 

During the interval until the next basic tape is made up and the 
new open-balance list printed, a compact auxiliary list is prepared. 
This is printed daily. This auxiliary list shows charges or credits which 
have occurred since the open-balance list was printed. The daily 
auxiliary list is cumulative. A customer inquiry can be answered by 
referring both to the open-balance list and to the auxiliary list. The 
latter constitutes about 10 per cent of all printing. 

This is a method for handling a file economically where a great 
quantity of information is involved if most items change infrequently. 
It is also being applied in many other types of installation. Telephone 
directories, certain inventories, stockholder lists, basic employee data, 
and many other types of lists are susceptible to this method of reducing 
the print-out and/or file-maintenance requirements. 

A bill register is prepared every time bills are issued. This contains 
everything that is on the basic record tape. Since changes occur to 
the "fixed" data, such as the customer address, a record-change list also 
is made up. This list is coded to refer to a daily record of register 
changes. When a particular item is consulted, either the bill register 
is correct or the changes can be traced quickly through the record- 
change indicator. 

Commonwealth Edison took delivery of an IBM 702 in the summer 
of 1955. They also had on order an IBM 705 for delivery in the sum- 
mer of 1956, and a second 705 a half year later. The first 705 was to 
replace the 702, the second to do general accounting. 

The programming group for the billing on the 702 installation was 
composed of five men from a public accounting firm, four from IBM, 
three from the Peoples Gas Light and Coke, and sixteen from Com- 
monwealth Edison. The latter were men who had averaged 24 years 
of service with the utility. The billing program required over 30 man- 
years to prepare. This does not include the time required for training. 

It was originally planned to take over a year to place the entire 
group of 1,800,000 accounts on the computer. An experimental group 
of less than 50,000 accounts was run first in duplicate to test out the 
electronic systems program. Once the "bugs" were worked out on the 
experimental group, conversion of the remaining accounts proceeded 
at an accelerating pace. 

Telephone Billing. Automatic Message Accounting is the name of 
a system used by the Bell Telephone companies. Also known as AMA, 
this system automatically records and processes on continuous paper 



68 Electronic Computers and Management Control 

tapes all data required for presenting charges on customer-dialed calls. 

Many telephone rate plans include "message-unit billing" for mes- 
sages to nearby points. Using the proper factor, relating the place 
called and the conversation time, this equipment computes the charge 
for each such message. These charges are then sorted and accumulated 
to provide a total message charge for each account. 

Where greater billing detail is required, the AMA system produces 
a record of each message, including such data as called telephone 
number, date, and time. These records may be either printed slips or 
punched cards, depending upon the type of subsequent accounting 
treatment selected to transform the AMA record into a customer's toll 
statement. This highly specialized telephone-message accounting sys- 
tem is notable for self-checking features which preclude overbilling. 

When you pick up your telephone and dial, a piece of equipment 
called a recorder perforates into a tape your telephone number, trunk 
number, and a message-unit rating code ( in the case of a message-unit 
call ) , or the called telephone number ( in the case of a toll call ) . After 
completing this entry the recorder is then free to record calls originat- 
ing from other customers. As soon as you receive an answer to your 
call the recorder is called in again. The answer time and trunk number 
are perforated. The machine is then free to continue recording other 
calls. When you hang up, the recorder is called in for a third time. 
It perforates the disconnect time. 

At 3:00 a.m. each recorder perforates a special pattern which will 
allow the tape to be cut. The tape is then forwarded to the accounting 
department for processing in one of their Automatic Message Account- 
ing centers. 

The accounting department splices the tapes together for each cen- 
tral office and then runs them through a series of five machines: 
the assembler, the computer, the sorter, the summarizer, and the 
printer. First, the assembler brings together the three elements of the 
call— the telephone number, the answer time, and the disconnect time. 
The computer computes the elapsed minutes of the call and converts 
the sum into message units for message-unit calls. The sorter sorts 
the messages into calling-telephone-number sequence. The summa- 
rizer adds the total message units used during the month for each 
telephone. Finally, the printer prints a list, intelligible to anyone, 
listing the telephone number and the total message units used during 
the month. It prints a slip containing details for toll messages. 



Studies and Applications of Electronic Systems 69 

Instead of using the printer, tapes for those central offices served by 
a revenue accounting office with punched-card message-unit or toll- 
billing procedures may be processed by a tape-to-card converter. This 
machine produces a punched card for each telephone number, repre- 
senting the total message units used during the month, and a card for 
each toll message. 

The output of the AMA center is then forwarded to the revenue 
accounting offices, where the message units are combined with mes- 
sage units for operator-ticketed calls and calls recorded by message 
registers. The message-unit allowance is then deducted, the charge 
and tax computed, and the result entered on the customer's bill. Toll 
messages must be rated in terms of money, recorded for intercompany 
settlement purposes, and entered on the toll statement. In cases where 
the revenue accounting office has punched-card equipment, the AMA 
output may be in the form of a card, and certain of the revenue account- 
ing operations are performed mechanically. 

Company studies have indicated the need for extensive analysis of 
all the various influences on the billing and receivable data processing. 
If, after a program has been prepared, too many exceptions must be 
inserted in the computer manually, then the benefits to be expected 
from rapid electronic processing may not be realized. 

General and Cost Accounting. A number of studies of electronic 
systems have come to the conclusion that there will be a substantial 
use of the systems in the fields of general accounting and of cost ac- 
counting. Among the characteristics which make this seem likely are 
the large volume of data handled in most accounting systems, the 
availability of cost standards according to fixed patterns, and the use- 
fulness of the information if it can be prepared rapidly. 

Perhaps the most ambitious of the general-accounting approaches is 
that of du Pont, which is testing both a UNIVAC and an IBM 705 on 
a tentative lease basis. The du Pont organization is studying the fol- 
lowing computer applications ( not listed in order of importance ) : 

1. Cost accounting. Production and cost records are required for 
80 plants. Two were selected for initial study. 

2. Payroll. Three plants and the needs of the 35,000 home office 
and other home employees were analyzed in initial studies. 

3. Production scheduling. 

4. Billing and sales. 



70 Electronic Computers and Management Control 

5. Personnel records. 

6. Property records. 

7. Accounts payable. 

8. Stockholder lists. 

9. Technical and scientific calculation. 

Du Pont studies began separately in several different functions. 
They were brought together under a steering committee in 1953. The 
first order for a computer was placed in 1954. The installation studies 
were expected to take at least two years. 

An installation group at Chesapeake and Ohio has been studying the 
possibilities of using electronics since 1952. They advocate what they 
call "a one-shot process." Using input from waybills, passing reports, 
wheel reports, interchange reports, car orders, rates and divisions, pay- 
roll data, materials requisitions, reorder levels, purchase orders and 
receipts, etc., they plan to produce as output: 

1. Timely digested results for management decisions and action. 

2. Exceptions for investigation. 

3. Detailed listings for reference. 

A company chart, used to explain this concept, says specifically, "The 
emphasis is on integrating the whole." 

In order to prepare for the data-processing system, C.&O. found it 
necessary to make careful studies of company needs and also to estab- 
lish a faster and more comprehensive communication system than the 
company already had. Both efforts yielded dividends even before the 
electronic installation. Methods research developed an application of 
statistical sampling to short-cut certain clerical procedures. Their 
work also has led to design of a new communication system, using 
teletypewriters and emphasizing improved use of the waybill. Com- 
pany spokesmen have characterized this improvement as "able to stand 
on its own feet in a highly competitive situation." 

An interesting example of an attempt to integrate an electronic sys- 
tem to provide for various accounting needs is the study made by Blue 
Shield. This study determined the way in which the RCA BIZMAC 
system could be used to handle enrollments and claims for the group 
medical plan. 

Three main programs were studied, each involving a different set of 
master tapes. 

The purpose of the first computer run was to update and correct the 



Studies and Applications of Electronic Systems 71 

enrollment master tape, to record separately the data removed from 
the enrollment master, and to compare the claims on a transaction 
tape with the enrollment master in order to separate those claims where 
the two disagreed as to certain data. 

The second run was to correct the experience master tape, to sepa- 
rate claims being processed into: 

1. Those having no prior related claims. 

2. Those having prior related claims. 

3. Rejected claims. 

Consecutive check numbers are assigned to those claims having no 
prior related claims. 

The third run was to update the doctor master tape, to prepare a 
tape for preparation of checks, and to prepare a tape for claim sta- 
tistics. 

The BIZMAC system features flexibility in sorting, conversion, and 
print-out, which gives it some advantages for this type of problem. 

Consolidated Edison, New York, has contracted for rental of two 
UNIVACS, and an IBM 705, for delivery in 1956. Two machines will 
be used on customer accounting and the third on general accounting. 
The decision to employ both types of equipment was made after a 
study of company requirements which took over a year. 

The company expects to process 1,500,000 customer accounts 
monthly, weekly payrolls for 23,000 employees, 9,000 other payrolls, 
100,000 inventory items of storeroom nature, and 9,000,000 property- 
account items. 

Monsanto Chemical has made a number of experimental cost analy- 
ses, using an IBM 701 and 702. They have reported that the results 
are encouraging, and that a program of further investigation, leading 
to use of later models for general accounting purposes, has been under- 
taken. 

Because of the size, intricacy, and importance of general and cost 
accounting systems, these applications have not been favored as one 
of the first for electronic computers. They probably cannot be handled 
economically by special-purpose systems. Extensive use of computers 
for accounting applications has had to wait for the introduction of more 
general-purpose data-processing systems. 

Among the uses which have been judged to be well adapted to the 
abilities of general-purpose electronic systems are the storage of 



72 Electronic Computers and Management Control 

standard cost information and automatic comparison with actuals to 
produce variance reports, the statistical analysis of variances to high- 
light only those which are worth executive attention, the development 
of data for cost estimates for bids and pricing through rapid assembly 
of individual costs stored in the system, and the development of costs 
for inventory and cost-of-sales reporting. 

Special-purpose Accounting System— Bank Checks. An example of 
a special-purpose electronic system for accounting purposes is the 
ERMA (Electronic Recording Machine, Accounting). This machine 
was sponsored by the Bank of America and developed by Stanford 
Research Institute. 

ERMA credits individual accounts with deposits and debits with- 
drawals. It maintains a record of all transactions and keeps a correct 
balance. It stores stop payments and hold orders and prevents over- 
drawing of accounts. It also sorts the checks. 

The machine has five input keyboards, each resembling a large 
adding machine. Different sets of keys are provided to represent 
the account number, the branch number, and the amount of check 
or deposit. 

Entries can be made at an average rate of about half a second. 
Switching between the entry keyboards is automatic, so that if several 
entries are made simultaneously, the machine will accept them all and 
only require a wait of about a second for the operator. 

The computer employs two large magnetic drums, providing stor- 
age sufficient for 32,000 accounts. 

Checks are precoded with magnetic ink, so that in the typical entry 
the only data that need be supplied by the operator is the amount 
of the check. The other information is placed on the back of the 
check as a set of magnetic marks. The name of the customer is also 
printed on the front of the check, so that a comparison may be made 
with the signature to be sure the customer is using the checkbook 
issued to him. 

The magnetic marks can be read at the rate of 1,000 characters a 
second. Dirt, ink, or other defacements do not interfere. The checks 
can be sorted at the rate of 10 a second. During test runs, errors were 
less than 1 in 100,000. 

In a typical entry, the operator depresses keys to indicate the 
amount of the check and the fact that it is a withdrawal. The ma- 
chine scans the back of the check and automatically depresses the 



Studies and Applications of Electronic Systems 73 

keys on the entire identification number. If the coding is not present, 
the operator can supply the number. 

The remainder of the operation is automatic. The machine obtains 
the current balance from the drum. It also scans the drum for pos- 
sible stop payments or holds. It subtracts holds from the balance 
before processing the check or notifies the operator of stop payment. 
Otherwise it merely subtracts the amount of the check and stores the 
information, including the new balance, back on the drum. 

The machine automatically sorts the daily activities and enters them 
on a master tape, once a day. On this tape the full record of all ac- 
count activity is stored in serial order. From this tape the monthly 
customer statements are prepared, as well as other information for 
management. 

The first ERMA was placed in operation in San Jose, California. 
The Bank of America contemplates use of 37 other machines at other 
branches. 

The research cost has been estimated as of the order of several mil- 
lion dollars. Neither the Bank of America nor Stanford Research plans 
to manufacture the equipment; rather it is expected that a major por- 
tion of the cost can be recovered by sale of the patents. It is expected 
that the machines can be produced in quantity to sell somewhere in 
the neighborhood of $200,000 to $300,000. At such a cost they will 
provide a saving over present practice, while relieving personnel of 
a tedious task. 

Summary — Accounting. The studies and installations discussed in- 
dicate the following advantages of electronic systems when applied to 
accounting: 

1. The computer does not have to replace the whole present account- 
ing system to reduce costs and be effective. 

2. In many cases some change is required in preparation of the 
source documents. 

3. The present group of skilled accountants and systems personnel 
often can be utilized after some training. 

4. It has been found that where a routine requires a large volume of 
either clerical or machine data processing, potential savings are 
sufficient to justify acquisition of a computer. Additional savings 
are then available from other applications. 

5. Computers are sufficiently reliable to be trusted with accounting 



74 Electronic Computers and Management Control 

routines which require a high degree of accuracy and depend- 
ability. 
6. Future savings can be achieved through anticipating potential 
uses of a computer, in current studies regularly made of account- 
ing systems and procedures. 

However, there are still some problems to be considered. 

1. Integration of the detail in accounting systems requires a sub- 
stantial period of time. There is a lack of personnel who can 
relate the accounting and electronic requirements. 

2. There is a serious risk of upsetting the accounting system. An 
installation must proceed cautiously, be checked and double 
checked at every point, both as to the electronic computer's pro- 
gram and as to the accounting system itself. 

3. There are internal and external limitations on the use of com- 
puters for accounting purposes. There are legal restrictions as 
to the type of record required for regulatory purposes and for 
admissibility as evidence in court. Original documents must 
still be retained for audit and tax reasons. 

4. To realize the savings may require the displacement of people. 
Resistance may be encountered from management as well as from 
unions. 

INVENTORY AND SALES 

Generally speaking, present inventory applications have been devel- 
oped by using the smaller electronic systems. By "smaller" is meant 
the computer systems in the hundred thousand dollar class, or less, as 
opposed to the million dollar machines. The latter do include inven- 
tory work among the operations they can perform. Moreover, they 
can tie it into the complete data-processing system. So far, however, 
this has not been accomplished extensively. 

There are several types of application which do not require that the 
handling and processing of inventory data be related directly with 
other accounting. Special electronic systems have been developed 
for these inventory applications. These systems record such informa- 
tion as incoming orders, or the status of a few thousand items of inven- 
tory. The cost has varied from thirty thousand to a quarter of a million 
dollars for the first models. 



Studies and Applications of Electronic Systems 75 

The savings which have been achieved with these smaller computers, 
or specialized systems, have been sufficient to justify a number of in- 
stallations. As a result, more special systems were installed for this 
purpose than for any other type of business information system, until 
general purpose computers began to be produced in quantity. 

In most branches of the retailing industry frequent physical inven- 
tory counts have been required. It has not been economical to keep 
perpetual inventory records, using present accounting systems. There- 
fore certain types of information, useful for management decisions, had 
to be provided by special groups of clerks. 

Experience has shown that these inventory information requirements 
could be supplied by special electronic systems. 

Careful studies had to be made before the machines were built. It 
has been particularly difficult to anticipate all future requirements. In 
the best systems as much flexibility has been built into the equipment 
as could be achieved economically. 

Application— John Plain. An operating example of an inventory 
system is the Speed Tally at John Plain. 

John Plain, a wholesale mail-order firm, distributes gifts and home- 
wares to thousands of customers. Not counting sizes and colors, they 
carry about 8,000 items in inventory. The business is highly season- 
able; daily orders can vary from 2,000 to 15,000, with an average of 10 
items per order. However, the majority of the orders are for one or 
two items. Most of their business comes in the fall quarter. 

Mr. Lachman, chairman of the board, initiated an investigation of 
electronics to solve John Plain's unit-inventory requirements. 

Mr. Richter, president, has stated: 

Filling orders promptly is the most important part of our business. De- 
lays and omissions are costly for us and disappointing to our customers. 
To ship promptly, we need balanced inventories. To keep inventories 
balanced, we need fresh, live information. This is especially true at the 
height of the season, when some items sell rapidly for a few weeks and 
then lie dormant until next year. . . . 

Our old system of keeping tally on incoming orders was acceptable, 
until a few years ago. We used the time-honored method of tallying in- 
coming orders by check marks, in 5's on prearranged sheets. When our 
business grew to the point where it took 40 clerks to tally and 20 more 
to accumulate the tallies, the system began to creak. Even with 60 opera- 
tors, we were able to supply only weekly reports to our buyers. They 



76 Electronic Computers and Management Control 

wanted daily reports! To give them that would have required a staff of 
150 tally clerks. 

A firm of consultants was hired. They recommended that the prob- 
lem be studied with the Engineering Research Associates (now a 
subsidiary of Remington Rand, a division of Sperry-Rand ) . The elec- 
tronic system built as a result of this study was called the Distributon 
by John Plain. Speed Tally is the name given to it by Remington 
Rand. 

The Distributon now requires only seven girls at most to keep a 
record of incoming orders. This compares with about 60 previously. 
Yet more information is available, more rapidly. 

The Distribution was given a fixed program. This enables John 
Plain to use a simple adding-machine type of keyboard as an input-and- 
control device. It requires a minimum of training of operators. When 
the operator depresses the keys for the quantity ordered and the item 
code number, the machine follows the program. It automatically cal- 
culates and records the new total. 

Nonexistent item code numbers are rejected automatically. A light 
flashes to indicate a mistake. If the operator punches a wrong item 
number she can correct the error. 

The program is set to enter automatically a quantity of one, the most 
frequent order size, merely by depressing the keys for the item code 
number. This saves the operator time and effort. The operators, as a 
group, can enter information into the system at an average rate of less 
than a half second per item. The machine takes information from one 
keyboard at a time. The other keyboards, which may have received 
a complete entry, are locked for a moment. Then their information 
is automatically accepted. In practice it was found that operators 
could prepare information faster than was expected, so that seven 
operators kept the "allotter" at capacity, although ten operators were 
originally planned to be used. 

Another part of the program enables any operator to read out, on 
a board, the total for any item at any time. This permits the operators 
to answer telephone requests by buyers for information on specific 
items. The total for any or all items can be printed out on adding- 
machine tape at the rate of 75 a minute. The items can be printed in 
any desired order. The order is controlled by a punched-paper tape. 

The machine has a number of features which give it some flexibility. 



Studies and Applications of Electronic Systems 77 

It is designed to utilize transistors when they become economical. 
The drum can hold 39,000 items. This gives it sufficient capacity to 
keep up with the growth of the firm. Or it can be used to store several 
kinds of information at the same time. 

The machine was not designed to provide accounting information, 
although the drum is a standard model and might be incorporated in 
a larger electronic system. It does not order or bill. This still is done 
by the buyer and the accounting department. 

After less than a year of operation, the John Plain tabulating group 
found that they could improve the operation of the machine by modify- 
ing the input keyboards so as to prevent the central unit from receiving 
certain types of erroneous entries. A check bit was added for each 
catalogue number to assure accurate transfer of information from one 
area to another on the drum. Another important change was to cover 
all relay chassis in the equipment so as to prevent dust from being a 
source of maintenance problems. 

Quoting Mr. Richter: 

... we find that it [the Distributon] has several accomplishments to 
its credit. With 10 operators [now seven] it does quite easily what would 
have required 150 tally clerks. It eliminated an employment problem, per- 
mitted us to engage more highly skilled operators, and saved much needed 
office space. More important, it has helped us to keep inventories balanced, 
with the consequent reductions of omissions and leftovers. But, best of 
all, we are giving top service to our customers. Some of these gains can be 
measured in dollars . . . but we believe that the dollars saved are out- 
weighed by the intangibles. 

The Distributon was built to suit the particular needs of the John 
Plain Company. The success of this machine does not mean that it 
is suitable for other companies. For example, another mail-order com- 
pany, Sears, Roebuck and Company, has studied the system. 

John Plain was fortunate in that its code was fairly complete. Sears, 
Roebuck employs a combination of numbers, letters, and description. 
This may impose problems of recoding the catalogue for machine 
purposes. 

Application— Department Stores. A less successful application of a 
special-purpose system was attempted by a large department store. 
After the machine was installed on the sales floor to provide unit in- 
ventory data, it was discovered that the capacity of the memory was 



78 Electronic Computers and Management Control 

insufficient for the data requirements. The machine had 14 digits 
available for each item, but the company soon found in practice that 
it needed at least 26 digits. Among other factors, additional digits 
were sought to provide the buyers with dollar information. 

The machine was an experimental model and not fully tested. As 
a consequence it had a number of operating failures. The manufacturer 
has estimated that it spent about $10,000 for special repair and 
maintenance on a machine that cost about $30,000 to build. (The 
machine was originally sold to the department store for considerably 
less than it ultimately cost to build and maintain. ) 

The department store has estimated that the machine saves the time 
of two clerks, which would have provided adequate pay-out at the 
store's original cost. But this saving would not be sufficient to justify 
an expenditure of over $30,000. This demonstrates the need for careful 
planning and cost estimation. 

A number of other department stores are experimenting with special- 
purpose electronic systems. Altaian's in New York has used a Magne- 
file with a storage capacity of 63 digits for each item. The machine 
can process data for about 8,000 items of furniture. 

The first 21 digits were used to identify 400 manufacturers and 100 
item classifications. 

The other 42 digits were devoted to seven groups of information, 
with six digits available for each: 

1. Dollars on order, at cost price. 

2. Dollars received in month, at cost price. 

3. Dollars received in month, at retail. 

4. Dollars in stock, at retail. 

5. Dollars sold for the period, at retail. 

6. Dollar returns for the period, at retail. 

7. Dollar markdowns for the period. 

Rich's in Atlanta is studying methods to control inventory. For this 
type of installation, one study has shown the following to be desirable: 

1. Number of items on order, per each item. 

2. Number in stock, per each item. 

3. Number sold for the period, per each item. 

4. Minimum reorder point, per each item. 



Studies and Applications of Electronic Systems 79 

In addition the system should be designed to group totals for many 
different classifications, to give sales figures, sales per salesperson, and 
totals on taxes, discounts, and salesmen's commissions. Separate totals 
show what amount of the sales were cash or charge, take or send. 

Rich's is actively investigating various electronic devices, and is co- 
operating with computer research at Georgia Tech for this pur- 
pose. 

Associated Merchandising Corporation, a federation of department 
stores, has a program of research in electronics. The general reaction 
expressed so far has been that the experiments have been worth the 
expense and show considerable promise. 

While special-purpose electronic systems show promise for dollar 
and unit control of inventory in department stores, there also have been 
a few studies of the application of general-purpose computers to this 
field. The advantage of a general-purpose computer is that it would 
permit integration of inventory information with ordering, billing, and 
other parts of the accounting system. Recognition of the need for an 
integrated system has long been recognized in the department-store 
field. A project known as Central Records was started at Kaufman's 
Department Store in Pittsburgh, about 1929. This system used pre- 
punched card inputs, one of which was given to the customer, one put 
on each piece of merchandise, and one given the salesperson. 

In the experimental runs all three tags were placed in the machine, 
which printed out the customer's name in the Central Records office. 
A clerk looked up the customer's credit and then pressed a button 
which signaled the selling floor so that the sale could be completed. 
There were standard card punches in the Central Records office, 
which took the information from the three cards in the transmitter, 
plus that from a cashier if cash was involved. When the transaction 
was recorded, the transmitter opened and released the cards. 

If the customer had lost her card, the information could be supplied 
by keyboard at the transmitter. Part of the merchandise card was 
given the customer as a receipt, and the customer signed a book kept 
beside the transmitter to recognize the sale. The punched cards could 
then be used for merchandise control, sales audit, and accounts-receiv- 
able posting. 

The system worked well enough for Kaufman executives to order 
further development, but machinery that could give satisfactory per- 



80 Electronic Computers and Management Control 

formance never was developed and delivered. Eventually the project 

was dropped. 

The Central Records concepts are currently being revived. 

Point-of-sales Recording. Sears, Roebuck has been experimenting 
with a punched-card system developed by Kimball and Potter Instru- 
ment. Under this system the inventory items are given a small, split- 
tag type of card. Part of the card can be detached and used to ini- 
tiate the accounting process. At first some difficulty was experienced 
in achieving accuracy in reading the tag. Tags became shopworn, 
bent, etc. The systems have gradually been improved, and installa- 
tions have now been made by a number of stores. 

Reading is done photoelectrically or mechanically. Information is 
usually transferred to a standard punched card, from which it is used 
in the normal way for various inventory and other purposes, but it 
could be given directly to the computer. The Kimball Company 
has withdrawn from the project to concentrate on other aspects of 
the computer field. Remington Rand has undertaken to develop this 
further with a machine called the Tag-omatic. 

A number of department stores, such as Macy's, and Robinson's in 
Los Angeles, are experimenting with point-of-sale recorders— systems of 
recording sales information at the selling desk. Telecomputing Corpo- 
ration and National Cash Register Company have devised methods 
for preparing computer input as the sale is recorded. Early models 
used punched-paper tape, prepared at each register and then trans- 
ported to computer input stations. 

A British concern, Rentrix Ltd., has developed a direct system, link- 
ing the cash registers, processor, and a "consulting unit" in the central 
office. The transfer of information can be made over distances by tele- 
phone wire. 

The cash registers can be fitted with keys to give as many categories 
of information as desired. The data are read out on a lighted screen 
at the central office, where the desired categories can be selected at 
Will. 

Dennison Manufacturing Company has a print-punch marking ma- 
chine which simultaneously records on price and inventory tickets all 
data that can be known about a unit of merchandise prior to sale. 
This is done by punching holes in the cards. Supplementary data, 
type of sale, salesperson number, price at which actually sold, etc., 
can be recorded at moment of sale. All data for each transaction can 



Studies and Applications of Electronic Systems 81 

be recorded on a tape. This may be teletyped to a home office for 
automatic processing. 

Airlines Reservisor. A special type of inventory problem is found 
in the transportation field. Space allocation is required for many forms 
of passenger operation. Reservations are made by ticket agents at 
scattered geographic locations. To obtain maximum loading it is 
desirable to provide the agents with information concerning current 
space availability for particular trips. It is advantageous to have this 
information while the prospective customer is still at the ticket-selling 
location. 

In 1952 American Airlines and the Teleregister Corporation intro- 
duced a special electronic system called the Magnetronic Reservisor. 
An input device is located at each reservation desk and at each of the 
major ticket-selling locations. The devices are connected by wire with 
a central computer located in New York City. Because of the speed 
of the system it can serve over 100 input locations. 

At each station the operator has a set of metal plates which have 
various indentations. The plates operate as a precoded programming 
device. Each plate represents a different flight or series of flights. 
The operator places the plate for the chosen flight in a slot in the input 
mechanism and depresses buttons for the number of seats desired and 
the date. 

A magnetic drum is used by the computer to store the number of 
spaces available for each flight. As the input requests are made, the 
computer automatically selects the proper flight and compares the sum 
of the old and new reservations with the total number available. If 
there are more seats available than requested, the computer sends 
back signals to the input device, which indicates that the reservation 
can be accepted by illuminating appropriate lamps. 

If the customer decides to take the space, the operator starts over. 
She flips a key on the input device, which automatically subtracts the 
number of seats reserved from the old total and stores the remaining 
inventory of available seats in the machine. A cancellation can be 
handled by the input device by adding to the total of reservations 
available. 

If the customer wishes to know the first flight which will be available 
to a certain destination, the operator can obtain this information. The 
whole operation takes only a fraction of a minute. If two stations try 
to reach the computer at the same time, only one will be connected. 



82 Electronic Computers and Management Control 

The other station will not receive an answering signal until the other 
has completed its entry. Experience has shown that such delays are 
not significant. 

The Reservisor has sufficient capacity for 1,000 daily flights, 10 days 
in advance. 

A desirable feature of any special-purpose system is that it should 
be capable of taking on additional duties. The American Airlines 
system is a good example. In the spring of 1955 the input devices and 
the central mechanism were altered so as to give information on ex- 
pected time of arrival of current flights. By pressing an added set of 
keys, but using the same plates to identify the flight, the operator can 
learn if the plane is on time, or how late it is, according to one of 
several time categories. 

The pay-out time is difficult to determine because of the intangible 
advantages. For clerical savings only it has been estimated as being 
as much as five years. The system has been in use for several years, 
22 hours a day, and has operated about 99.8 per cent of its scheduled 
time. 

United Airlines has installed a Reservisor system, and several rail- 
roads have installation programs. 

Stock Quotations. An interesting variation of this inventory tech- 
nique, also developed by Teleregister Corporation, is the Toronto Stock 
Exchange Bid-Asked Register. This is a magnetic-drum machine 
which makes prices available to 200 dial request subscribers, at a rate 
up to 4% requests per second. 

Prices telephoned by reporters on the Toronto Stock Exchange trad- 
ing floor are entered by keyboard operators and transmitted to Tele- 
register display boards above each trading post. The prices also ap- 
pear on a large board before the operators, to enable them to check 
accuracy. 

A drum provides for storage of 2,000 sets of prices. (Six hundred 
was the initial usage. ) 

To request a quotation, a broker dials a three-digit number. At the 
central office in the Exchange building this number is decoded, par- 
tially by the older Teleregister equipment, and finally by the new 
electronic equipment. The desired stock is automatically located on 
the drum. A mechanical transmitter sends the prices out. The answer, 
including stock number and prices, appears on a ticker tape at the 
requesting station. 



Studies and Applications of Electronic Systems 83 

Combining Special- and General-purpose Machines. The problem 
of relating special systems (such as file maintenance, or inventory- 
count machines) to general-purpose computers has been explored in 
a number of studies. Some firms require a large volume of memory 
only for certain parts of their operations. A satisfactory solution may 
be to have a special-purpose and a medium-sized general-purpose 
computer. Large-scale general-purpose equipment may not be eco- 
nomical. Rapid selection of data regarding any one of a large number 
of items is expensive in most general-purpose computers, so this would 
be handled by the special-purpose equipment. The less extensive 
operations then could be put on a medium-sized computer. 

On the other hand, companies which purchase special-purpose equip- 
ment to satisfy special needs are confronted by a serious problem. 
The advantages of their special electronic data-processing equipment 
often may be made obsolete by changes or improvements in the com- 
pany's information flow. Changes in business procedures also may 
render obsolete the special-purpose equipment. 

In order to avoid such obsolescence, the electronic equipment manu- 
facturer is faced with the problem of designing some flexibility into 
his special equipment, or he may persuade the user to make a special 
study of requirements in order to convert existing procedures to ones 
which are likely to persist and which the equipment can handle eco- 
nomically. The business user is then faced with the difficult task of 
evaluating changed operating methods and of deciding whether they 
would result in a saving. 

If electronic equipment is sufficiently flexible, an installation can 
be made evolutionary rather than revolutionary. The equipment can 
be adapted to a company's procedures, both as they now exist and after 
they are changed. General-purpose computers provide this program- 
ming flexibility. They can be adapted to many (but not all) proce- 
dural changes. They also permit more than one function in the organi- 
zation to use the equipment. On the other hand, increase in flexibility 
usually increases cost. 

To gain the advantages of both types of systems some electronic 
manufacturers, as well as some individual companies, are developing 
special-purpose computers which can enter inventory information, 
make preliminary calculations, do preliminary summarizing and sort- 
ing, and make elementary decisions. The general-purpose computer 
would receive from the special computer only that information which 



84 Electronic Computers and Management Control 

is necessary to post the inventory accounts for general accounting 

purposes. 

Mail-order System. This "combination" approach is of particular 
interest in connection with problems like those of the Spiegel Corpo- 
ration. Spiegel has circulated a detailed technical description of its 
operations with the request that electronic manufacturers suggest pos- 
sible systems. 

In a typical mail-order house incoming orders are processed through 
item tally, then cash analyzing or credit, as appropriate. Then they 
proceed to a ticket-pulling operation. Tickets, preprinted with appro- 
priate information, such as price, weight, and description of the item, 
are available by catalogue number in racks. A clerk processes a given 
customer order by selecting the appropriate ticket for each catalogue 
number on the order. 

The tickets are then manually marked for color, size, catalogue by 
which purchased, and quantity. Mailing labels are prepared. The 
orders are scheduled and stamped with a packer number. This sched- 
ules the people who will wrap the orders for mailing. 

The tickets are then separated from the customer orders, sorted by 
the various merchandise departments, and dispatched. At each de- 
partment they are tallied by size, color, and catalogue number, flagged 
if out of stock, then released for stock pulling. Each merchandise item 
is placed, together with its pull ticket, on a conveyor-belt system for 
transportation to the wrappers. 

Customer orders are separately routed to the packer. The packer 
number determines the time and the packer station which is to receive 
the merchandise and order. The complete order is merged and packed 
at a packing station. It is then weighed, stamped, the customer is 
billed, and the merchandise is shipped. 

Data gathered at the tally stations serve to prepare management 
reports on stock conditions, sales by items, sales by catalogue, etc. 

Receiving and transfer information is routed to the tally stations 
to correct their inventory figure. The same data also are used to pre- 
pare reports on purchases, receivals, and shortages. 

The tally operation often proves troublesome. Especially during 
seasonal rushes it induces error into the system. Studies have shown 
that file-maintenance equipment, as described previously, can help. 

Following this it is planned to automatize many of the clerical 
operations. For example, the ticket-pulling and pread justing functions 



Studies and Applications of Electronic Systems 85 

can be replaced by electric typewriters directly connected to a special- 
purpose inventory computer. A typewritten form replaces the pull 
ticket. As the clerk who receives the order types the catalogue num- 
ber, the data is accepted by the special computer, which refers to its 
memory, looking up description, weight, price, and other information. 
These it sends back to be typed automatically on the form. The 
computer also checks available inventory and the total of requests for 
the item. If stock is available, the machine indicates that the item 
can be sold. It then adjusts the inventory and the tally of orders. 
Canceled orders are maintained in the same manner. If the item is 
out of stock, the computer indicates this. At the same time it keeps a 
tally of orders for out-of-stock items for buyer information. 

The special-purpose computer also prints out invoices and other 
associated documents for completing the order. Such a special-pur- 
pose computer also would be designed so that managers could deter- 
mine the status of any item at a moment's notice. 

Periodically, the contents of memory can be transferred to a mag- 
netic tape, in an ordered form. When combined with other informa- 
tion, this tape can be used by a general-purpose computer to produce 
management reports, postings to the accounting records, etc. 

By using both special-purpose and general-purpose computers the 
sorting problem is minimized, especially if data are collected in a 
preset fixed order. In some systems separate tapes could be produced 
for each requirement, at the same time, by the general-purpose com- 
puter. Each tape then could be transferred to a printer, in any 
sequence. 

Large-scale Inventory System — Air Force Supply. The Air Force 
has under development an extensive inventory-control system. The 
first installation is an IBM 702 at the Oklahoma City depot. 

The system is designed to keep a complete central-inventory record 
by processing all transfers to the system, plus those between the 
depots and bases and between the bases and the unit supplies. All 
requisitions would create a punched card, which could be transferred 
over the new IBM Transceiver system. The Air Force has leased 
lines 24 hours a day to build up a communications network for this 
purpose. 

A major study project is under way to evalue the four large-scale 
commercial computers— the UNIVAC series, the IBM 700 series, the 
BIZMAC, and the RAYCOM. It is expected that eventually several 



86 Electronic Computers and Management Control 

dozen of these large-scale general-purpose machines will be utilized 
for this purpose. 

One of the advantages of the system is to permit the Air Force to 
keep larger supplies near the operating level. Better control would 
facilitate interbase transfers of scarce items in a manner not now pos- 
sible. Also it would enable the system to build some flexibility into 
reorder points and standard order amounts. These could be made 
to depend upon the quantity of an item available in the over-all system. 

Summary — Inventory and Sales Applications.' The studies and in- 
stallations discussed, concerning inventory and sales applications, in- 
dicate the following advantages of electronic systems: 

1. Special electronic systems are economical for some inventory and 
sales problems. 

2. To obtain economical advantages it is necessary to integrate the 
data information needs and the design of the electronic system. 
The results of this integration have been simplified input, suffi- 
cient machine capacity, and special features which reduce operat- 
ing costs. 

3. Inventory needs can be combined with production, purchasing, 
and accounting requirements through the use of special-purpose 
and general-purpose electronic computers. 

4. It is possible to utilize input-output devices, at decentralized loca- 
tions, that are directly linked to a central computer. 

However, the following should also be considered in the application 
of electronic systems to inventory and sales: 

1. A special system developed for a particular application in one 
company may not be usable by another company in the same 
way. A single different factor may be sufficient to make it 
uneconomical. 

2. Special electronic systems still have limitations. The amount of 
rapid random-access storage that can be obtained at reasonable 
cost is not large enough for many applications. ( Manufacturers 
are working hard to improve this.) 

3. Obsolescence of special-purpose equipment is also a serious con- 
sideration. 



Studies and Applications of Electronic Systems 87 

PRODUCTION CONTROL AND SCHEDULING 

During the years after World War II it has become increasingly 
apparent to management that methods of production control and 
scheduling need improvement. Developments in production control 
and scheduling have taken two directions— mechanization and intro- 
duction of new techniques. 

Some companies have moved toward mechanization by using 
punched-card machines. Such applications often strain the ability 
of the equipment. Many manual operations are still required. Use 
of punched cards is limited by the amount of information that can be 
processed economically. Also, it is difficult to put in sequence certain 
operations on punched-card machines. 

A number of applications are contemplated which relate linear 
programming and other new developments with production control 
and production scheduling. Current research in these areas has indi- 
cated that production control and production scheduling are different 
types of problems. Each requires a different approach. Solution of 
production-control problems requires the use of mathematical tech- 
niques which maximize profits, taking into consideration such factors 
as the current forecast of sales, the available resources of the company, 
the operation costs, and the state of technology. The solution provides 
a program which sets forth the best product mix for a given period. 

Scheduling, on the other hand, is the detailed guidance of operations 
in conformance with the program. The schedule sets forth the times 
when materials, parts, labor, and machines are to be made available 
to specific operating units; the amounts of each resource to be pro- 
vided; and the expected results. Somewhat different mathematical 
techniques are useful. 

However, both production control and production scheduling can 
benefit from the use of high-speed computing equipment. 

There have been few actual applications of electronic computers to 
production control and scheduling. Extensive work is being done by 
research groups in the larger corporations, and it is generally believed 
that the area of production control and scheduling is one of the most 
promising for applications of electronic computers. A large volume of 
computations is required. There is need for faster preparation of 
schedules in order to achieve better control of material, labor, and 
machine utilization. Clerical handling of data is expensive. 



88 Electronic Computers and Management Control 

A computer can do the necessary work efficiently and economically. 
Moreover, the computing need not be done at the plant. Several 
companies have declared that they intend to place a computer in a 
central location to serve all of their operating plants. Others, however, 
are decentralizing such work; many have ordered a medium-sized 
computer for each plant. 

Recent research indicates that better mathematical methods for 
production control and scheduling may soon be available. Dr. A. 
Vazsonyi has formulated mathematically a method for determining 
parts requirements. Work on mathematical techniques to solve pro- 
duction problems is also being done at the RAND Corporation, UCLA, 
Carnegie Tech, and by the Cowles Commission at Yale University. 

A number of companies are making scientific studies of production 
control and scheduling. Among these are General Electric, Lock- 
heed, Ramo-Wooldridge, Thompson Products, Ordnance Supply of the 
United States Army, SKF Industries, Standard Oil of New Jersey, 
Hughes Aircraft Company, International Business Machines Corpo- 
ration, and Litton Industries. In most cases these companies plan 
to use computers to perform the necessary computations and to prepare 
the paper work. 

Application— Lockheed. A group working at Lockheed has pre- 
pared full sets of production orders on an IBM 701 computer. They 
have turned out, in a few hours of computer time, aircraft orders 
which otherwise took more than a week to prepare. The manager of 
production scheduling has stated that use of the computer has not 
caused the discharge of a single man and is not likely to do so. He 
has emphasized also that use of the computer has enabled Lockheed 
to produce more realistic schedules and programs, which result in 
lower cost of production. 

Lockheed has found that the higher speed of the IBM 701 and 704 
scientific series is so desirable for their production needs that it offsets 
other advantages of the IBM 702 and 705 business series. Production 
scheduling requires a fair amount of calculations. The company has 
contracted for several 704's, but also ordered a 705 for general account- 
ing purposes. 

Lockheed managers hope to use their computers to help them 
analyze the problem of production acceleration. The company went 
through two periods of rapid acceleration— during World War II 
and during the Korean conflict. Partly because of scheduling difficul- 



Studies and Applications of Electronic Systems 89 

ties, the company sometimes experienced increasing costs as the num- 
ber of planes built increased, rather than achieving a reduction in cost 
per unit, as should be expected. The managers plan to analyze past 
data, taking into account the labor market, skills required, rate of pay 
within the community, availability of space, housing and transporta- 
tion, as well as a host of in-plant factors. By use of multiple correla- 
tion techniques they believe that factors can be developed for use 
in future programming and scheduling of production. The factors can 
be combined into "hours per pound," and used to estimate the time re- 
quired to build airplanes. Use of a computer also facilitates con- 
tinuous reexamination of the factors used for the estimates as produc- 
tion proceeds. 

Underwood Study. The Electronic Computer Division of Under- 
wood has examined the production-control system of the company's 
Hartford works. The Hartford plant produces typewriters. The study 
was made to determine which parts of the production-scheduling sys- 
tem could be handled economically by the Elecom 125. 

The proposed electronic data-processing system was designed to 
continue use of all the source documents currently used. As experi- 
ence was gained with the computer, various record forms could be 
dropped, particularly as the production-control personnel gained con- 
fidence in the computer. It was extremely dangerous to change old 
flows of documents without careful investigation, because the docu- 
ments might be needed for purposes other than those immediately 
evident. 

The problem of change-over to a new electronic system was examined 
in detail. It was concluded that the change-over could be accom- 
plished either at one time or piecemeal. 

A comparison between the present system and the one using the 
Elecom 125 is shown in the table on page 90. 

Underwood expected to save the time of these people while using 
the computer only 4 hours per day. The extra machine time could be 
used on other applications, such as payroll. In addition to savings in 
clerical costs, the company felt sure that human errors in the present 
system would be minimized by use of more highly trained personnel. 
All entries would be made twice and checked by one of several alterna- 
tive methods. Mathematical errors would be held to a minimum. 

The system would prepare the operating reports. A daily production 
report would be provided, one which compared the actual production 



90 



Electronic Computers and Management Control 





Present system 


New system 


Function 


How often 
prepared 


Number of 
personnel 


How often 
prepared 


Number of 
personnel 


1. Master schedule, kept up 
to date every day 


Daily 
Daily 
Daily 
Once every 

2 weeks 
Once every 

3 months 


12 
6 
6 

12 

7 


Daily 
Daily 
Daily 
Every 2 

weeks 
Once every 

3 months 


2 


2. Stores and ordering record . . 

3. Special-parts schedule 

4. Production-manufacturing 
schedule 

5. Material-allotment 
summary 


2 
2 
2 

3 



to that on the master schedule. A control report would be prepared 
daily to show the parts-inventory status of those items which were 
in a plus or minus balance. Feedback of information would be pro- 
vided only for those items which required management attention. 

Other Studies. Systems for production control and scheduling can 
be usefully interrelated with the data processing for accounting, inven- 
tory, sales purchasing, management control, and other parts of the 
business system. It seems evident that a general-purpose computer 
will be desirable for most installations. In order for an electronic 
system to be able to handle a variety of operating problems or varying 
methods of data analysis, it must be flexible. To achieve this flexibility 
in a computer with a built-in-program is likely to be so expensive that 
the user might as well obtain a general-purpose computer. However, 
file-maintenance computers may have applications in scheduling, for 
inventory control and for priority choice. 

An example of the complex interrelationships which exist between 
production control and the other areas mentioned is found in the 
experience of the Bureau of Aeronautics. 

The Bureau of Aeronautics has used a CRC 107 for production 
scheduling of government-furnished equipment to meet production 
schedules of aircraft manufacturers. The scheduling requires the 
monthly quantities of items needed. It involves adjusting schedules 
to provide installation lead times to meet airframe company require- 
ments. Adjustments to monthly requirements are needed because 



Studies and Applications of Electronic Systems 91 

of changes in types and quantities of aircraft and equipment. Changes 
occur daily; at time of initial studies they averaged about fifty a week. 
Schedules are completely reviewed each month. 

One of the problems encountered by the Bureau was the lack of 
understanding between the person familiar with the problem and 
the person programming it. This difficulty will be a common one for 
similar applications in the future. One reason for this difficulty was 
that the individuals presenting a problem have been so misinformed 
as to what computers will do that they do not present their problem 
in sequence, in simple terms, and in complete detail. As a result, 
the programmer had to go to several persons to gather the entire in- 
formation. Of course, in many cases it is nearly impossible for any 
one person to present the problem in its entirety. Some problems are 
too far-reaching in scope, and too many people are involved. 

There is a feeling among some production people working in job 
shops that the production-control problem probably is too difficult for 
an electronic machine or system to handle. They suggest that elec- 
tronics may be useful only in production control of standard large-lot 
process-type industries, like an automobile assembly plant. However, 
some research workers believe that pay-off in production control 
eventually will be found in any situations where present manual or 
punched-card methods are taxed to capacity. 

Production Control—Study of Job Shop. An in-plant study of pro- 
duction control was undertaken by the Management Sciences Research 
Project at the University of California, Los Angeles. It was sponsored 
by the Logistics Branch, Office of Naval Research. The study was 
headed by Richard Canning. 

The plant studied was a large job shop in the local area. It employs 
about 1,000 persons and supplies complex metal assemblies to the air- 
craft industry. 

The production-control manager had tested various methods to con- 
trol the flow of work through the machine shop, including punched- 
card systems and modern adaptations of the Gantt-chart technique. 
He had chosen a relatively efficient manual system based upon the 
use of control boards in each fabricating department. 

The major problem confronting the production-control manager is 
to estimate the work load on each machine-tool section a few days 
in advance. Since there is no standard routing, jobs can and do flow 
into a machine-tool section from all the other sections. Bottlenecks 



92 Electronic Computers and Management Control 

develop overnight. When such bottlenecks occur, considerable ex- 
pediting action is required. Since bottlenecks cannot be predicted, 
life in this plant is a continual process of expediting for production 
management. 

The proposed electronic system records the present location of all 
orders within the plant. It can compare this location with the sched- 
uled location and select for management action those orders which 
fall behind schedule more than a specified amount. 

The requirements for parts can be determined by the machine. 
Parts shortages can be detected for expediting action. 

Intermediate start dates can be computed from information con- 
cerning the present location, priorities, and the flow times for future 
operations. For expediting action estimated due dates can be com- 
pared with desired due dates. 

In addition to control functions, the proposed system can perform 
much of the clerical paper work. The maintenance of perpetual in- 
ventory records for each part number, maintenance of customer order 
information, issuing of shop orders and purchase orders, and many 
other functions are within the scope of the system. 

A special-purpose scheduling computer to be loaded from the shop- 
order-status tape was suggested. The computer starts working its 
way into the future, hour by hour. When a machine tool is shown 
to be available, the scheduling computer scans through all waiting 
shop orders and picks the one with the highest priority that is slated 
to go on that type of machine tool. At any desired time, the com- 
puter can stop and total up the number of shop orders waiting in 
each department. This gives a picture of scheduled hours versus 
available hours. 

A control board is used to present a picture of the progress of critical 
shop orders. A route sheet overlay is placed on the board. Lights 
indicate the position of the shop orders. As the scheduling computer 
works its way into the future, the lights "step across" the control board. 
At the same time, a display panel indicates the "hypothetical future 
date." 

Should one of the critical shop orders not make satisfactory progress, 
or should a future bottleneck become apparent, the chief expediter has 
several choices for corrective action. He can change priorities to 
move some jobs faster. He can schedule overtime work, or he can 
send certain jobs outside. By manipulating in this way, he can 



Studies and Applications of Electronic Systems 93 

derive a satisfactory schedule for the next week or two. Also, he can 
get a rough idea of the future by letting the computer run out a month 
or two in advance. 

A rough estimate of the time scale for this operation is 15 minutes of 
electronic computer time per 40 hours of shop time. 

The system designed was only of a preliminary nature. The purpose 
was to see if it was economical and to set a pattern for subsequent 
research. Later, however, several companies in the Los Angeles area 
used a similar approach for studies of their requirements for elec- 
tronic equipment to handle production control. 

Summary — Production Control and Scheduling. It is generally 
recognized that the area of production control and scheduling offers 
many opportunities for the use of electronic data-processing equip- 
ment. Nevertheless, the problems are complex. Some experts believe 
that new devices are required. Others feel that the problems must be 
formulated in mathematical terms, using new techniques of analysis. 

In general, it is expected that successful applications will require 
the talents of a team of experts who can design or appraise new 
equipment, who understand production-scheduling problems in de- 
tail, and who can apply new methods to achieve solutions. 

DOCUMENT HANDLING 

In addition to the use of electronic systems for business-data proc- 
essing, there has been a considerable amount of effort directed toward 
solution of the problem of handling documents. 

Regardless of the efficiency which electronic communication may 
achieve, paper documents will continue to be used for many purposes, 
at least in the foreseeable future. The sorting and delivery of paper 
records and authorizations is one of the most costly of certain types 
of clerical operations. 

An example is the volume of checks which must be processed by 
banks. Several banking groups, and individual banks, such as the 
Bank of America, have been actively investigating this field. They have 
sponsored the development of print readers and of methods of coding 
by attaching metallic dots or magnetizable strips to the documents. 

The major problem is to achieve initial recognition of the identifying 
and quantitative information on the document. Once that is done, 
sorting, accumulating totals, and transferring of the document can be 



94 Electronic Computers and Management Control 

done automatically. For further processing, the information can be 

placed on media suitable for use in computers. 

Development of the necessary equipment has reached a stage where 
commercial installations are feasible. 

SUMMARY 

Applications of electronic computers have now been made to a 
number of different data-processing functions. The degree of initial 
success has varied. Substantial savings in time and in clerical effort 
have resulted. Cost savings, however, have been more difficult to 
achieve. It is believed that in many cases the economic benefits will 
come from the ability to perform more service at practically no in- 
crease in costs, rather than to cut costs dramatically by performing 
the same operations electronically. 

Study of various types of installations shows that there are certain 
aspects which are unique to each installation and a few which most 
systems seem to hold in common. The specific ways and times at 
which information is required for planning and operating are peculiar 
to each situation and must be studied in detail. These needs, together 
with different physical circumstances at points where events occur 
and are recorded, result in specific requirements for input devices, and 
to a lesser extent for special processing and output equipment. 

Similarities exist which enable one to group applications in sev- 
eral general categories. Some applications, usually of a scientific or 
engineering type, require moderate input and output but large 
amounts of computation. Some, like payroll, require voluminous in- 
put and output and in addition require that records for almost every 
"item" (e.g., employee) be changed at each processing. Others, like 
inventory control, require voluminous input and output, relatively little 
computation, and reference only to a fraction of the items in the 
master file in any one processing run. The latter case is sometimes 
called a "low collation ratio." 

Each type of situation calls for a different type of electronic sys- 
tem. The first is usually called a "scientific computer." The second- 
such as for payroll— is the "general-purpose computer." The third, 
and in some respects the latest to be developed, is the "file computer." 

Finally, there are some aspects that seem fairly general. Of these, 
the most important is the basic concept of the flow of data through 



Studies and Applications of Electronic Systems 95 

typical electronic systems. Very frequently this flow is an elabora- 
tion of Figure 4, which is a highly condensed "flow chart." 

PAPER KEYBOA RD 

J<? «? 9 

\ J / 



INPUT TAPES 






\ 1 / 



MERGE .SORT 



PROCESSED INPUT TAPE 



INSTRUCTION 
PROGRAM 




MASTER RECORD TAPE 




PROCESSOR 



PROCESSED MASTER RECORD 
TAPE FOR SUBSEQUENT USE 



PRINT OUT IF DESIRED 




EXCEPTION TAPE OR 
SPECIAL REPORT TAPE 



PRINT OUT IF DESIRED 



Figure 4 



Some of these steps may be omitted— the input may go directly 
from the keyboard to the processor, or sorting may not be necessary. 
On the other hand, the system may be greatly elaborated— there may 
be a number of reports prepared simultaneously by the processor. 
However, the general pattern is already evident in many types of ap- 
plication. 



96 Electronic Computers and Management Control 

In virtually all cases company experience has encouraged executives 
to expand their efforts in this field. There is a general belief that 
in large and many medium-sized companies electronic data processing 
eventually will dominate record keeping, communication, and pro- 
vision of managerial information. 

Experience has also shown that achievement of the results which 
are ultimately possible will not come easily. The administrative 
problems will be tremendous during the period of adoption of the 
new systems. New methods of system analysis, emphasizing the 
scientific approach to business problems, will have to be employed if 
the electronic machines are to operate with maximum effectiveness. 
In the longer range, an entirely new and different approach to man- 
agement planning and control is indicated. And, ultimately, auto- 
mation of the office will be coordinated with automation of the factory 
and with the use of the computer as a tool for scientific research if 
the full benefits of electronics in business are to be achieved. 



CHAPTER FIVE 



Administrative Problems Experienced in Introducing 
Computer Systems 



The introduction of an electronic system creates a number of ad- 
ministrative problems. Old equipment must be replaced in an 
orderly manner. People must be retrained, regrouped, or relieved. 
Various kinds of experts may have to be hired, familiarized with the 
company's problems, and fitted into the organization. Data-gathering, 
-processing, and -reporting methods will have to be revised. 

Coordination of the change in machines, men, and methods will 
require a high order of administrative ability, based upon an under- 
standing of the interrelationship of various aspects of management 
planning and control, electronic systems, and integrated information 
systems. 

General Electric's Installation. There are no installations as yet 
which can serve as models for a solution of all these problems. The 
General Electric Company has been the first in America to experi- 
ment with the installation of a large-scale general-purpose business 
computer. Their experience is being watched in the hope that it will 
eventually provide answers to many of these problems. 

The General Electric Company built a plant at Louisville, Ken- 
tucky, to relocate and replace the facilities for the production of five 
major household appliances. This modernization and geographical 
centralization of certain lines gave the company an opportunity to 
experiment with and to develop newer methods of production and 
of data processing. In a new operation, there is less concern about 
replacement of machines, retraining of personnel, and introduction 
of new systems. Executive attitudes, therefore, favored investiga- 
tion of automation and electronic data-processing systems. 

97 



98 Electronic Computers and Management Control 

Approval of the directors was secured for a study of electronic sys- 
tems. Of course, authorization of capital expenditures for an actual 
installation was contingent on the results of the study. 

In companies where the support of top management has not been 
obtained, groups investigating the feasibility of computers have been 
hampered in defining problems which should be investigated. The 
investigations have been restricted to the functional areas under con- 
trol of those who take responsibility for the study. As a result, al- 
though some of the studies have shown that an electronic system can 
be used for a particular application, it could not be shown that it 
was the best application, or even that it was feasible from the over-all 
company point of view. 

For example, in one company the controller found an application to 
cost accounting which would reduce the clerical costs substantially. 
He had his staff group make a detailed analysis of the way the com- 
puter should be used. Meanwhile, with the approval of management, 
the production department was engaged in an automation program 
to reduce manufacturing costs. The system the production group 
proposed would, in their opinion, make most of the data handling 
unnecessary. Months after the two studies had started the overlap 
was finally recognized. By then both groups had strong vested in- 
terests in their own programs. As too often happens, a choice between 
the competing interests was made in a political environment. 

The General Electric Company placed their study under the di- 
rection of a systems and procedures group. They decided to obtain 
the help of experts in this new area. They retained a firm of certified 
public accountants who, since 1948, had been studying the possibility 
of application of electronics to business systems. Various electronics 
manufacturers were also invited to participate. 

The first problem they encountered was a set of misconceptions 
which prevented some divisional managers from giving electronics 
serious consideration. Most of these dealt with the high cost of the 
equipment and its supposed lack of ability to fit into business opera- 
tions. However, study showed that a break-even point could be 
reached when a computer was used as little as 2 hours a day on four 
clerical operations. 

Savings were computed on the basis of salaries, space rentals, and 
equipment depreciation applicable to those clerical jobs eliminated 
in four limited routine applications. No value was given to such in- 



Administrative Problems in Introducing Computer Systems 99 

tangibles as more prompt reporting, the ability of management to make 
more informed decisions quicker, and possible reduction of invest- 
ment in inventories. It was believed that had these important factors 
been included, a break-even point could have been reached with fewer 
applications and less computer time. 

As a result, approval of the directors was obtained for an electronic 
installation. 

Soon after the decision was made to go ahead with an electronic 
computer, the Louisville team decided that the operating management 
of the five appliacance departments needed indoctrination in the use 
of computers. This was especially important if the program of mod- 
ernization and centralization of Louisville was to proceed smoothly. 
Since 1941 General Electric had been following a policy of expanding 
its productive capacity by building small, geographically decentralized 
plants. In addition, the company had been actively pursuing a 
program of decentralization of management responsibilities. The 
data-handling and -processing function, up to this point, had been 
under the direct administration of each of the department managers. 
This meant each could and did develop his own control and report- 
ing system. Only the general accounting system was under the func- 
tional supervision of the central home-office accounting staff, for policy 
purposes. The introduction of the electronic computer would dis- 
rupt this system of relationships to some extent. 

Early meetings with the operating management stressed the fact 
that there would be no major realignment of responsibilities. The 
computer group would serve only as a service center. Analysis of 
reports and actions resulting therefrom would still be the responsibility 
of departmental managers. The meetings served to provide the man- 
agers with a forum to ask questions and to work out their individual 
and mutually shared misgivings and problems. 

The same degree of understanding was required of the management 
and staff reporting to the department heads. It is at this level that 
the actual working out of the details takes place. Over a period of 
time the foremen and the various staff members had developed close 
relations with decentralized data-processing groups. The prospective 
loss of this personal contact could give rise to misgivings. Under 
the old system, fast feedback of data was possible. When a figure was 
questioned, a call or personal visit could quickly clear up the issue. 
The same degree of personal rapport would not, among other things, 



100 Electronic Computers and Management Control 

be possible with a centralized electronic system. This could lead to 
questioning of the accuracy of the data reported. 

Decentralized data processing had afforded a maximum of flexibil- 
ity. Requests of the operating management for particular data or 
reports could be met with a minimum of explanation as to what was 
wanted by the managers. 

Centralized data processing requires careful scheduling and setting 
of priorities. Requests for particular data or special reports call for 
detailed explanations, which were not required when the managers 
and the data processors worked so closely together. 

The group discussed potential computer applications with execu- 
tives from all functions of the business— financial, marketing, purchas- 
ing, the manufacturing groups, and so on. 

Points stressed at these meetings included: 

1. No change need be made in existing data-collection procedures 
or in the reports. 

2. All original documents would still be returned to the originating 
department. 

3. Clerical-cost savings would result. 

4. The managers would be able to receive additional informa- 
tion to assist them in performing their responsibilities more effec- 
tively. 

Immediate and short-range applications of the computer which 
were discussed included: 

1. Data needs for payroll. 

2. Material scheduling and inventory control. 

3. Order service and billing. 

4. General and cost accounting. 

Longer-range projects for the computer were also considered. 
Feasible future applications included: 

1. To prepare monthly, quarterly, and annual budgets. Various 
suggested combinations of product mix would be used to prepare 
complete budgets and forecasts, employing the previous week's 
information on labor and material cost variances. 

2. To improve the work-load balance on automatic-laundry as- 
sembly lines. Each of these calculations required 50 hours of 
desk work, whereas the computer could prepare one in a few 



Administrative Problems in Introducing Computer Systems 101 

seconds. It would be practical to prepare several hundred pos- 
sible solutions and choose the best one. When production rates 
changed, new balances could be prepared for immediate use. 

3. To handle sales and inventory reports. Later still this could be 
extended to include initial phases of short-range and long-range 
projections and forecasts, including multiple correlations with 
economic and other factors affecting the sale of products. 

4. To prepare factory machine-loading schedules. 

5. As a later possibility, to coordinate materials, machine loading, 
and labor loading for production control. 

6. Eventually to do long-range forecast of sales, a forecast employ- 
ing such data as birth rates, family formations, disposable income, 
the level of employment, availability of electricity in homes, and 
comparisons of competitors' models and prices. 

7. Eventually also to establish an integrated system which would 
provide dynamic distribution analysis. It would provide for 
management an up-to-the-minute picture of the retail trade and 
of distribution pipelines. The purpose of this integrated system 
would be to enable management to adjust selling effort and 
factory effort to make maximum use of resources for more profit. 

As a result of these discussions, it was mutually agreed to under- 
take the short-range projects in the following order: 

1. Payroll. 

2. Material scheduling— inventory control. 

3. Order service and billing. 

4. General and cost accounting. 

The procedures group determined to select equipment whose cost 
was fairly certain to be recovered, while at the same time choosing 
equipment able to handle more extensive and potentially more profit- 
able applications in the future. For this reason search was limited to 
large-scale general-purpose computers. As they were eager to start 
the program, they leased the only large-scale machine readily avail- 
able at that time, the UNIVAC. 

Lyons, Ltd., a restaurant chain in London, England, had a similar 
experience, except that no commercial machine was available at the 
time they became interested. They hired several electricians and radio 
technicians, none with computer experience, and in less than two years 
developed a working electronic system. The system now prepares 



102 Electronic Computers and Management Control 

weekly payrolls for over 30,000 employees. Special devices were de- 
signed to check the accuracy of the input data and to insert new 
employee names into the proper sequence in the master file. They 
are planning to use the computer for sales analysis and other purposes. 

Because of the interest which was created by their installation, 
Lyons has expressed willingness to develop similar equipment for 
other companies. 

General Electric organized an installation group to prepare for the 
computer. The primary source of readily available personnel for this 
group was the business-procedures section of the company. This 
section had men: 

1. Who were familiar with the operation of the business. 

2. Who had training in previous systems and procedures. 

3. Who had at least a college diploma and, in many cases, a 
master's degree in business administration. 

Psychological tests were used to choose men from this group. The 
tests were designed to select personnel with high intelligence and 
ability to think logically and reason abstractly. 

Several members of a certified public accounting firm were re- 
tained to provide expert help in the system installation. A course 
was started to train people in programming the UNIVAC. The train- 
ing was done at the manufacturer's school in New York City. By fall 
in 1953, about a half-dozen programmers were sufficiently trained so 
that they could start on the actual programming and coding of the 
payroll routines. 

The build-up of the programmers continued. By using the facilities 
of the manufacturer the cost of training was kept down. Retaining 
the certified public accountants to aid in the programming made it 
possible to minimize relocation of personnel once the total program 
was coded and installed on the UNIVAC. 

General Electric's approach was to schedule, as a first application, 
a program which called for great accuracy and which at the same 
time would relieve the staffs of the various operating departments 
of considerable routine clerical computation; thus it would provide 
savings substantial enough to be evident to all levels of management. 
If the computer handled the payroll accurately, it would alleviate 
some of the apprehension as to accuracy of the system. At the same 



Administrative Problems in Introducing Computer Systems 103 
time, reduction of the cost of routine clerical computation was an 
objective which operating executives could appreciate and share. 

However, original recording of time and production data remained 
with the operating departments. This substantiated the policy that 
the only change would be in the processing. This was also necessary 
because the computer-installation group reported directly to the wash- 
ing-machine department. 

A third benefit of the application chosen was in the extent of the 
service which the computer could provide. It was estimated that ap- 
proximately one hundred different types of reports would be prepared 
for all levels of management. These reports would be prepared from 
the same output tapes as were used for payroll calculation, rapidly 
and inexpensively. 

Originally it had been hoped that the first running of the payroll 
could be done about May, 1954. Before this was feasible, however, 
it was necessary for the manufacturer to deliver a high-speed printer. 
The printer was not available until summer, and other difficulties, 
primarily mopping-up details, slowed conversion so that the first trial 
run of payroll, in parallel with the actual payroll, was not achieved 
until several months later. The first solo run occurred in early fall 
of 1954. 

It is difficult to assess all the facets of this immensely interesting 
pioneer effort. The group at Louisville has been reconstituted, with 
leaders receiving promotions elsewhere. Company executives have in- 
dicated that there may be somewhat less emphasis in the future on 
further clerical routines as early applications, and greater emphasis on 
the types of applications which were considered "secondary" in original 
planning— the use of the computer on mathematical planning to 
achieve better production planning and better marketing analysis. 
It is still believed that the clerical applications are feasible, but that 
greater pay-offs await in the other areas. 

It is expected that, on the average, initial applications will occupy 
the electronic system less than one shift. The remaining time will be 
available for further applications and for mathematical work. 

General Electric did not place production control and scheduling on 
the schedule of applications which would be among the first to be 
placed on the computer. Instead, they set up a special group to study 
the problems which must be solved before a successful computer ap- 



104 Electronic Computers and Management Control 

plication of production control and scheduling can be made. This 
permitted the company to hire the necessary experts, not available 
within the company, and to integrate them into the organization with 
a minimum of disturbance. This group of experts also reported to the 
washing-machine department. 

The same group can be used to work on problems connected with 
other future applications. Among those contemplated is the develop- 
ment of a method for sales forecasting that can be related to the 
programming and scheduling of production, purchasing, sales effort, 
and financial needs. 

A by-product of the investigation is the discovery that it is feasible 
to link the data systems for several plants that are geographically 
separated. The company has considered the preparation of payroll 
for the Erie, Pennsylvania, plant on the Louisville UNIVAC. The data 
can be transferred rapidly and economically, either through input 
media or by teletype. 

New tools, such as new methods of mathematical analysis, were 
found to have many implications for the proper organization of an 
electronic data-processing system. In addition, the analysis has 
provided economic benefits even before computer application. Dr. 
Melvin Salveson, formerly of the UCLA Management Science Project, 
has reported an improvement in production control which saved sev- 
eral hundred thousand dollars. 

Organization Problems. No pattern has been evolved as to where 
the computer group should be placed in the company organization. 
Various arrangements have been tried. In some companies the com- 
puter groups report to the controller, in some to the chief accountant, 
and in others to the heads of purchasing, production, sales, or produc- 
tion control; in a few cases the computer groups report directly to 
the president. Some large corporations have more than one group. 
For example, one has a group responsible to the accounting function, 
and another group in the production department located in a different 
city. Another large company has assigned to different divisions the 
responsibility for experimenting with different types of equipment. A 
large merchandising corporation has followed a similar approach, as- 
signing different input and storage devices to different stores, where 
they are installed in different types of selling departments. 

The existence of several computer groups within an organization, 



Administrative Problems in Introducing Computer Systems 105 

while offering certain experimental advantages, has already led to 
some internal organizational conflicts. Difficulties arise because each 
of the groups tends to compete in several ways— e.g., for skilled per- 
sonnel in an extremely tight labor market, for the attention of top 
management, and so on. Each group tries to prove that its approach 
is the best, and it is difficult for top management to determine the 
degree to which each should be encouraged. Different approaches 
are desirable if, for example, one group is interested in clerical-cost 
reduction and another is primarily concerned with analysis and pro- 
duction control. But in many cases management finds it hard to judge 
these experiments because of lack of experience in dealing with 
this new field. 

In some companies each group has been permitted to choose the 
equipment it feels is best, with the result that machines from different 
manufacturers have been chosen. So far, manufacturers have tried 
to discourage firms from assembling an electronic system from several 
sources; some equipment has been designed deliberately to make it 
incompatible with other manufacturers' machines. If, for example, 
input media are not compatible ( some machines read "five-hole paper 
tape," some read "seven-hole paper tape"), then the two groups can- 
not assist each other to balance loads or to substitute during break- 
downs, without expensive rehandling of the input media. The coding 
and programs may be different, so that even if the media are com- 
patible rehandling may be necessary. 

If the different systems are ever to be integrated, then additional in- 
vestment for conversion equipment will be required. Programs may 
have to be redone. This is a major reason for emphasis on integrated 
systems, as in the U.S. Steel, SKF, and Chesapeake & Ohio installations. 

Programming and Personnel Problems. It is generally believed that 
it is necessary for an executive to have some understanding of how 
electronic systems work in order to be able to appreciate the personnel 
problems which are involved. In large measure the reason for this 
belief is that new systems will require the services of personnel trained 
in new techniques of method analysis, and of programming, coding, 
and operating the new equipment. 

One expert has gone so far as to declare that programmers may 
very well represent a new category emerging in society. A pro- 
grammer must be a meticulous thinker. Such concepts as "reason- 



106 Electronic Computers and Management Control 

able" or "obvious" or "comparable" are not satisfactory in his work. 
He must operate with a set of definite criteria, with all the relation- 
ships spelled out. 

Herbert Grosch, of General Electric, one of the pioneers in ad- 
ministering a computer group, has declared that a programmer must 
be a good chess player. Or, failing that, "he must be a person who 
would play a good game of chess if he played chess!" 

The significance of this lies in the fact that a programmer must 
be a person not only with an analytical and logical mind, but one who 
has a keen sense of spatial relationships. He needs these in order 
to visualize the ordering of the various elements in a program and 
to place the instructions and information in the various storage de- 
vices of the computer so that handling of items will proceed to 
best advantage. 

In one comprehensive installation study, for example, it was found 
that the computer's facilities were always too limited to do all work 
in one pass every day. Therefore the first step was to draw up a 
chart which showed the top-level operations for the computer— the 
way that tapes flow from one job to another. Input and output steps 
were also charted. 

However, in order to do this it was necessary to make assumptions 
as to the number of tape input units available. If it were found that 
twelve were required at one time for a major job, this meant that an- 
other major job should not be charted to use six if it could be more 
economically set up to use twelve (once it was assumed that twelve 
would be available anyhow). Moreover, it was necessary to assume 
that magnetic drums would or would not be available, and that cer- 
tain file arrangements and codes would be used, etc. 

These determinations are part of the study that should be under- 
taken before the precise nature of the electronic system is specified. 
Then, after installation, when specific flow charts are made up, the 
first assumptions have to be refined. Possible changes may be sug- 
gested which might benefit one program but hinder another. 

Program policies must be established. These ensure uniformity 
with regard to such things as tests to determine whether a tape can 
safely be used for writing (writing erases material already on the 
tape), or whether the tape holds the list of accounts receivable, which 
might as well be kept. 

From matters such as these it should be evident that handling a 



Administrative Problems in Introducing Computer Systems 107 

computer calls for abilities which are not precisely like those re- 
quired for any activity heretofore present in a typical business. Even 
punched-card operations are considerably different; it cannot be as- 
sumed that a capable man from the punched-card department will 
necessarily prove adept at handling the new tasks. Many such men 
are studying to make the transition, and it appears as though on the 
average they will perform capably. However, experience is limited, 
and so far there are few tests with which to separate quickly the 
mediocre from the good. 

This is important because there is some evidence that in the end 
extremely capable men may be worth many times as much as those 
who are just fairly good. In several cases a good program has been 
found which cut operating time in half over a fairly good one. But 
there is no way of determining these matters until it has been done. 
And there is no assurance that some other program may not be much 
better. Research in the area of generating routines, pseudo codes, 
etc.— attempts to program the computer so it will aid in its own 
programming— indicate that companies which can develop a top- 
notch programming staff may have a tremendous competitive ad- 
vantage. 

Programming Experience. Experience with large-scale general- 
purpose computers has shown that programming an application usually 
is time-consuming and expensive. In some instances the original 
expectations have been exceeded by over 100 per cent. 

Testing the first programs for the computer is also expensive. The 
workers must proceed carefully; it is rare to have a program work 
out right on the first try. Frequently it has been found advisable to 
have the computer installed a month or several months before it is 
running full time on actual work, so that programs can be tested. This 
means that rental of previously installed equipment must continue 
during the interim. Most large-scale computers have been run in 
parallel with older equipment for at least two full routines (such as 
payroll) before change-over. 

Preparation of a single program takes from a few weeks to as long 
as 6 months, not counting time for testing. In one case, delivery of 
a computer was postponed for several months, at request of the cus- 
tomer, because the programs were not ready. 

So far manufacturers have been generous in their assignment of 
skilled programmers to assist installation groups. Even so, there is a 



108 Electronic Computers and Management Control 

shortage of skilled programmers. It has been found that it takes 
about 3 months or more to teach a person to handle a large-scale 
computer properly and as much as 6 months to train him to program. 
Even then there is no assurance that he will be "skilled" in the sense 
that has been given to that term. 

More programmers are being trained all the time. However, 
whether sufficient will become ready is still a question. At least 10 
to 15 programmers and analysts are needed on large-scale installa- 
tions; some firms talk of hiring many more. The smaller installations, 
using one or more of the many medium-sized computers, will require 
at least 3 to 5 programmers. If orders for large-scale computers and 
for medium-sized computers hold firm, there will be a need for sev- 
eral thousand skilled programmers by the end of 1956. Companies 
are attempting to anticipate their needs by sending men to the pro- 
gramming schools now being run by computer manufacturers and by 
some colleges, but the problem may grow acute. 

As a result of the scarcity, salaries for good programmers are in the 
five-figure class. This has considerable significance for the continuing 
costs of running installations with large groups of programmers and 
the other skilled operators who are required. 

A number of companies have found it worthwhile to start pro- 
gramming their computer a full year before they get their electronic 
equipment. They have rented a few hours on machines in order to 
pretest the programs, when necessary. Meanwhile, the programmers 
are getting a thorough training, both in the abilities of the machine 
and the requirements of the business. 

Equipment Problems. Delivery of computers, and especially of 
the first models of newly developed devices, sometimes has been 
delayed. In one case an important output device was delivered 4 
months late. In another, a new device proved unreliable. Such delays 
can add to the cost of a computer installation, even though partly 
borne by the manufacturer. In some cases they were not contemplated 
in the original savings studies. 

Reliability of well-tested components of electronic computers has 
not been of much concern to users. In one case a company reported 
that, over a period of more than a year, more than a hundred en- 
gineering changes, some major, were made in the computer, but these 
changes rarely had caused delay or other concern to the operating 
group. 



Administrative Problems in Introducing Computer Systems 109 

Newer devices, particularly for input and output, have been subject 
to a much greater incidence of breakdown. In one installation a new 
output device was down over 25 per cent of scheduled operating time. 

In the case of some medium-sized machines the "de-bugging" time 
for the equipment has been relatively short— often a few days have 
been sufficient. Some larger installations have taken a matter of 
months. 

Future installation studies should benefit from experiences such as 
those outlined above. Although in some cases this means that sav- 
ings expectations may not be quite as great as some previously visual- 
ized, nevertheless the conclusion of most companies has been that 
they should have started their studies earlier and more intensively 
than they did, for the savings from the installations are still sub- 
stantial. 

On the other hand, it can be seen that total costs will be large, 
after including such other items as space requirements, need for air 
conditioning, possible personnel displacement, crowded conditions 
during initial parallel operation, and so on. Failure to install the 
right system could be a major mistake. Nor can it be assumed that 
because a computer is "general purpose" it can handle all types of 
data processing efficiently. For example, in some installations it has 
been found that the high-speed storage capacity of the machine, or 
of i elated devices, is not sufficient to store all the programming steps 
that the operators would wish for maximum efficiency in running a 
particular routine. This has meant, in some cases, that a program 
must be broken into as many as fifty, a hundred, and even more sec- 
tions. Since each section can run only that part of the data for which 
it is suited, this means that the data must be sorted before being 
placed on the input or that all the data must be run several times— 
either of which is an inefficient use of the computer, as a rule. In 
some cases the presorting requirement has raised the time required for 
an application by several hundred per cent over original hopes. 

Improved design of the computer or of programs is not the whole 
answer to this problem. In some cases it will be advantageous for a 
company to revise its data-handling and accounting requirements to 
put them more in line with electronic capabilities. This problem has 
received inadequate attention from accountants. 

However, as indicated in Chapter 3, the logical design of electronic 
systems can be expected to change so that the machines will become 



110 Electronic Computers and Management Control 

less a revised version of scientific computation centers and more an 
expression of the much different business computation and processing 
requirements. Such design improvements eventually may reduce the 
need for skilled programmers and yet make the machines more 
flexible and cheaper. 

In addition to problems related to the central equipment, there 
are also some difficulties in connection with the media which can be 
used for initial recording. Some machines still rely heavily on punched 
cards, and some place emphasis more on magnetic tapes. Experience 
has shown that there are some significant advantages to using magnetic 
tapes: information does not get lost in handling as frequently as with 
punched cards, since information cannot easily be left out or altered. 
The inputs may be converted into related sets of answer tapes. These 
answer tapes show the results of computations, so the situations before 
and after processing are both available. This record is not now avail- 
able with machines using only punched cards, as a rule, unless con- 
siderable print-out of intermediate stages is attempted. 

Another advantage is that errors can be edited more easily, when 
discovered. In a number of installations an error on about one 
punched card in a thousand is customary, requiring extensive trouble 
shooting later. The tape error often can be handled by the com- 
puter if impossible codes, addresses, or information are involved. In 
some cases the computer can be programmed to make a reasonable 
assumption and go ahead, keeping a record of the assumptions made 
for later verification. 

On the other hand, there are some advantages to punched cards. 
Tape markings are invisible. To print out a full tape in order to 
locate trouble is time-consuming, so that the operators often try to 
avoid this. They theorize as to possible errors and attempt solutions 
by altering the tape or the program. In one installation the manage- 
ment has insisted that a print-out be made whenever an error on the 
tape is suspected. With punched cards, on the other hand, the file 
can be more quickly searched to find the item causing the difficulty. 

Present speed advantages of tapes as media may tend to be 
reduced in the near future. New devices reading a thousand cards a 
minute are under development, and with punched-card electronic 
computers they may offer more of the same advantages as tapes. Sort- 
ing of card information can be made much more rapid. Tape sorting 



Administrative Problems in Introducing Computer Systems 111 

is quite time-consuming at present, though special devices to improve 
this are also being developed. 

Tapes are cheaper than cards, per item of information stored. 
Still, they may cost $50 to $100 each, and though reusable, several 
installations are using many hundreds of tapes, so the investment is 
not inconsequential. Occasionally defective tapes have somehow 
passed inspection, so that time is lost until the defect is discovered 
during operations. Foldovers or dust may cause items of information 
to become lost. It is necessary to handle tapes with care and to 
keep installations exceptionally clean. 

Air conditioning is another problem with large-scale machines, 
which at present develop considerable heat from their large numbers 
of tubes. Space requirements and strength of flooring, while not 
usually critical, must be considered. Power requirements of elec- 
tronic computers may require special wiring. In some installations 
these costs were not considered sufficiently in the original estimates. 

Sequence of Applications. Another problem of major proportions 
for operating managers has been to select the best sequence of ap- 
plications for the electronic system. This problem has been aggravated 
by the fact that some computer groups have been unable to estimate 
savings accurately. Underestimates of time required and overesti- 
mates of possible savings have resulted in pressures from manage- 
ment which in turn have sometimes influenced the choice of ap- 
plication. 

There is some difference of opinion as to the type of application 
that should be attempted first. Should it be one which is likely to jus- 
tify the cost of the installation quickly? Or should it be one which 
avoids deadlines, minimizes contacts with customers or employees, and 
avoids undue reliance on a new device? In the latter case the savings 
may not be nearly as great. 

On the other hand, many groups believe that it is in this area of 
analysis that greatest benefits will be derived, not from clerical savings. 

In some cases the first applications have demanded great accuracy, 
with strict time schedules, etc. In such applications the effects of 
mistakes or mishaps during initial runs may be quite serious. Such 
errors in the past sometimes have led companies to remove various 
types of nonelectronic machine installations, even though application 
groups believed the installations eventually could have been made 



112 Electronic Computers and Management Control 

economical. One large company reached the following conclusions 
concerning their large-scale installation: 

1. The first concern is to get the computer on a paying basis. 

2. Benefits from the first application should be tangible and clearly 
defined. 

3. Considerable savings may be available through use of the 
computer in areas not involving large clerical operations. This 
is particularly true of untried areas where adequate equipment 
for handling large volumes of complex calculations was lacking. 

4. Attempts to handle such areas should be made only after com- 
petence has been acquired through experience with more fa- 
miliar procedures. 

In another company, virtually the opposite of each of these con- 
clusions has been reached. 

Internal Auditing Problems. As yet most internal auditors and pub- 
lic accountants have not expressed much concern about the difficulty 
of auditing electronically prepared reports. In fact, at various meet- 
ings, speakers have reassured the audiences that basic accounting 
principles will not be changed and that most auditing techniques will 
still apply. 

There is not much experience from which to draw conclusions as 
to the accuracy of such statements. It does seem as though some 
problems may arise. For example, a precept of internal control of 
long standing has been "separation of duties," so that considerable col- 
lusion would be required to falsify records. But the number of 
humanly handled steps will be drastically reduced by some of the new 
electronic systems. 

A number of years ago, an embezzler, immortalized by St. Clair 
McKelway in The New Yorker as the "Wily Wilby," made a practice 
of returning to his office on weekends and rerunning an entire set 
of punched cards through to new ledger pages. He managed to pass 
audit, in spite of pocketings amounting to several hundred thousand 
dollars (not counting the small bonus the firm gave him for good 
work ) . 

Almost by definition the new electronic system experts will have 
to know how to run their whole system. 

It appears obvious that close control of input tapes, programs, and 
output records will have to be maintained, with as much separation 



Administrative Problems in Introducing Computer Systems 113 

of duties as possible. If there is any slip-up, a single expert might 
be able to rerun a tremendous volume of transactions, by old standards, 
in only a few hours. 

A major safeguard against this in most present installations is the 
fact that the expert would not benefit from such falsification, since 
he does not have access to disposable assets or to means for mis- 
directing checks. 

Auditors are at present talking of reliance on the same original 
evidences— the papers from which input tapes are now prepared. 
Outside confirmation is expected to help. However, there is going to 
be considerable pressure from installation groups to eliminate as much 
as possible of these original papers, since if they could be prepared 
automatically, by print readers, by voice recorders, by counters, by 
direct inputs from cash registers and time clocks, etc., a substantial 
saving in clerical work would result. 

Moreover, even though no fraud is present, there will be difficulty in 
following through complicated programs of instructions to see that 
proper allocations are made— that cost figures represent the intent of 
the policies set by executives. The machines will process so much in- 
formation, and so fast, that there will be veritable avalanches of 
data to be studied, unless the controls can be properly evaluated 
without such detailed investigations. 

In their capacity of passing on the efficiency of the data-processing 
system and the efficacy of the internal controls, some groups, both 
of internal and of public auditors, have come to the conclusion that 
they will have to learn the details of operation of these new electronic 
systems to a far greater degree than others have thought necessary. 
(Or, perhaps, than they would like, in view of the time and effort re- 
quired to understand the systems.) While it is true that basic ac- 
counting principles may not be changed by the advent of electronics, 
yet these workers realize that one must comprehend a data-processing 
system sufficiently to choose the proper check points, if he is to judge 
whether or not accounting principles are being applied properly and 
whether or not the internal controls are adequate. 

Entirely changed concepts of cooperation between the internal 
auditors and the operating managers may be in the making. 

Research Nature of Computer Installations. Some managements 
have, perhaps unconsciously, placed heavy pressures on their com- 
puter groups to show some quick, tangible results of their efforts. 



114 Electronic Computers and Management Control 

The executives appear to fail to understand that computer applica- 
tions are applied research projects in every sense of the term. Long 
study periods are necessary in order to define, measure, and evaluate 
the business-data requirements before they can be adapted success- 
fully to the computer. 

During the initial study phase, few concrete results are visible. 
Members of the group spend their time talking, reading, and— pref- 
erably— thinking. Only a brief report may be forthcoming at this 
stage. At best, a few sets of program instructions will be prepared, 
a few hundred pages of paper. 

Yet if this stage is not handled properly, the later periods will 
suffer. Already there have been instances where special-purpose 
systems were inadequately designed, as a result of failure to make 
sufficient preliminary investigation. In one company, for example, 
an inventory application was installed which perpetuated three dif- 
ferent inventory-numbering systems. Revision of the numbering sys- 
tem is a major task, but once accomplished it will result in benefits 
that are also of major proportions, both in better operating systems, 
with unnecessary steps and duplications eliminated, and also in 
more efficient utilization of the electronic equipment. Moreover, 
if the company ever tries to integrate its information systems, the 
costly change-over will have to be made anyhow. 

Summary. Transition to use of electronic computers has created 
a number of administrative problems. The cost of the installation is 
substantial. Selection of proper equipment and competent personnel 
is not a simple matter. Experience has shown that preliminary esti- 
mates frequently are less accurate than for other types of resource 
allocation. 

Conversion of present routines, without improvement or integra- 
tion, seldom results in the most efficient installation. Therefore a 
detailed study of the business requirements must be made. These 
requirements must then be related to the capabilities of electronic 
equipment. 

Presently available equipment is not always as satisfactory as 
some experts believe it should be. Personnel able to make analyses, 
prepare programs, etc., are scarce. Training of company employees 
is usually desirable but does not necessarily result in the required 
competence. 

Provision of better data for decision making is considered to 



Administrative Problems in Introducing Computer Systems 115 

offer one of the most promising areas for future electronic applica- 
tions. However, before this can be accomplished in full meas- 
ure, a company usually will have to adopt new methods of analysis. 
The whole approach to management planning and control may have 
to be revised. 

Chapters 6 to 10 will discuss these analysis problems. Their rela- 
tionship to the selection of the most suitable electronic system for a 
business then can be presented. 



CHAPTER SIX 



Management and the Scientific Approach 



Experience has shown that electronic systems can reduce clerical 
costs. However, in many instances the real significance of the use 
of electronic equipment lies elsewhere. Even if an installation in- 
creases costs, it may well be worthwhile. 

Electronics can serve as a vehicle for the company to study, re- 
organize, and improve not only the business-data processing and com- 
munication systems but also the system of management planning and 
control. For this reason a number of companies now believe that 
they should revise their previous approach to installation studies. 

As is evident from the data presented in Chapters 4 and 5, when a 
company begins to study a computer application, its executives face 
two general types of problems, which may be characterized as tech- 
nical and administrative. With regard to technical problems, the 
following conclusions are fairly well established: 

1. The ideal electronic computer for data processing is not yet 
available. 

Present computers are not efficient for handling the complete 
accounting function, production scheduling and control, and 
all of management reporting. Electronic systems do not even 
readily debit and credit in the accepted accounting sense (al- 
though it must be admitted that some experts argue that this 
is no longer necessary). 

2. After an application has been placed on a computer, it usually 
becomes evident that the machine can do more than that particu- 
lar function. 

For example, a computer organized to handle payroll usually 
can be programmed to do cost accounting for labor, labor report- 
ing, etc. However, as more and more functions are added, it 

116 



Management and the Scientific Approach 117 

also becomes apparent that the computer has some limitations. 
These may result from voluminous input and output or the 
need for storage. Or, extensive sorting may be required. In 
some cases the current system does not lend itself readily to 
application on the computer. Either the system must be changed 
in order to take advantage of the computer, or else new com- 
ponents must be added to the electronic processor in order to 
handle the business needs. 

3. While the programming requirements of the computer can be 
met by training the present clerical staff, a few weeks or months 
are not enough to prepare them to make efficient use of the 
computer. 

Inexperienced workers often program many redundant op- 
erations. As new data requirements appear, company program- 
mers may find that they have not had sufficient training to know 
how to adapt the machine to the new problem. And there is no 
assurance that even with years of experience all the men will 
become highly proficient on all types of problems. 

This is not solely the fault of the personnel. Some companies 
have included computer engineers in their analysis groups; their 
assignment is to invent new devices to help get easier solutions 
to the changing problems of the company. 

4. Whenever anyone prepares a program for a computer, there is 
always the danger that some important data are being omitted. 

The omission of an item may not be discovered until later, 
perhaps too late for efficient revision. The problem has also 
been evident whenever mechanical systems were installed, but 
it was easier to correct, simply by adding personnel or equip- 
ment. 

It is possible for a computer program to be checked out much 
more effectively when solving a mathematical formula than when 
dealing with a business procedure. A mathematical formula is 
a complete system within itself, one that can be checked for 
consistency and completeness before and after solution. 

No formula exists today which can be used to show a com- 
plete, integrated business system. Each company has its own 
procedures. In almost all cases the systems are largely the 
result of a series of semirelated historical accretions. 

More important, over the past ten years there has been a 



118 Electronic Computers and Management Control 

growing need for data for management planning and control. 
This need has not been satisfied by the accounting system, 
the production-control system, or by the cost system. Even 
if the latter were correctly programmed, problems would con- 
tinue to exist. 

These technical problems have their counterparts in what might 
be called "administrative" problems: 

1. Executives may be disillusioned in the early stages of application. 

Expected savings may not be realized. If an installation is 
predicated upon successful handling of one routine, and if the 
electronic system design was oriented to that one routine, it 
may be difficult to correct or to improve the situation. For 
example, it takes so long to make an installation for payroll or for 
order billing that, by time the system is in operation, many pre- 
mature decisions may be reached as to the capabilities of the 
machine. 

In one case, after two years' work installing a system, the 
savings achieved offered no pay-out of the original cost of equip- 
ment and installation. For this reason, some company executives 
are impeding further utilization of the computer within their 
organization, even though what they have learned makes it 
highly probable that further applications would be more re- 
warding. 

This administrative problem is complicated by the fact that 
first applications often are chosen to show quick savings so that 
management will be satisfied. Large areas which in the long 
run offer more substantial profit are neglected because the com- 
puter application group fears that management will call the 
gains intangible. These areas of potential benefit often involve 
the method of organization and the responsibilities of manage- 
ment, and so cannot be approached without the firm support of 
top executives. 

2. Interfunctional conflicts may arise. 

The integration of business systems of necessity affects all 
groups in the organization. Centralization of data processing is 
often required in order to justify a large general-purpose elec- 
tronic system. When a company is in the process of decentraliza- 
tion, or has just completed a move toward that direction, this 



Management and the Scientific Approach 119 

need for centralization of data processing cannot help but create 
administrative problems. 

Conflicts often have existed in various functions anyhow. For 
example, animosity between accountants and production en- 
gineers over cost allocations has become almost proverbial. 
The integration of accounting and production in such situations 
leads to organizational conflicts. Where engineering needs for 
computation are substantial, the engineers may be able to buy 
their own computer. If the conflict is deep-seated, in some com- 
panies this has resulted in the purchase of two different types 
of computers. In other companies, where this duplication has 
been prevented, one group usually tries to take over the other. 

3. As the move grows for placing new functions on the computer, 
a need for specialists arises. 

Company personnel seldom have the background adequate to 
meet all the new requirements in a short period of time. Elec- 
tronic engineers are needed to maintain the equipment. Skilled 
programmers are required to code the computer. 

In addition a new group of specialists are required— men who 
understand the tools of "management science" or, as it is often 
called, "operations research." Such specialists today are in 
short supply. One company advertised and searched for over 
half a year in an attempt to hire one. The cost of such a man 
generally is 20 to 50 per cent higher than that of specialists al- 
ready working in the company. 

4. Unless there is clear evidence of the need, executives tend to 
distrust and oppose large-scale organizational changes. 

Most persons who have been closely associated with the in- 
stallation of computers believe that a limited approach is not 
satisfactory. Morale of the installation group suffers when they 
see their efforts blunted by enforced limitations. Yet execu- 
tives are rightly reluctant to undertake substantial reorganiza- 
tion until it is clearly desirable. 

Some experts have circumvented opposition by setting their 
subordinates to work on a particular application while they them- 
selves, and some of their newly acquired specialists, work on the 
more significant problems. In one company the head of the com- 
puter application group has been running the second shift on the 
approved application so that he and his specialists could have the 



120 Electronic Computers and Management Control 

first shift available in which to experiment with applications closer 
to management planning. Of course, top management was not 
aware of this, and was not expected to learn of it until proposals 
could be worked out and documented. 

5. Communication is poor between installation groups and man- 
agement. 

To establish an integrated system is a major accomplishment. 
Those who have tried this have found that the language they use 
is not familiar or readily translatable to management. The new- 
ness of the field creates a difficulty in itself. 

6. The scheduling of organization changes is a difficult problem. 

Once made aware of the desirability of using an electronic sys- 
tem, management soon realizes that the changes incident upon 
installation will require careful planning. Alteration of the whole 
communication system of a company is not a minor matter. 

At the same time, executives are confronted by other develop- 
ments, such as atomic energy and automation. A priority prob- 
lem exists. To what areas should management give primary at- 
tention? Should it be the computer, or the other areas? When 
should these various changes be put into effect? 

The advent of electronic computers, of automation, and of 
atomic energy has imposed upon today's executive an unpre- 
cedented need for developing an orderly process of introducing 
change, while maintaining a stable organization that can return 
a profit to the owners. 

A survey shows that more and more computer groups are adopting 
a scientific approach to these management problems. This is a 
natural development. The most effective use of electronic systems 
requires: 

1. Definition of the business problem. 

2. Measurement of the important quantitative factors in the situa- 
tion. 

3. Expression of the relationships between the factors. 

4. Formal recombination of the factors. 

5. Testing and improvement of the resulting program. 

This is a scientific approach. 



Management and the Scientific Approach 121 

HISTORICAL DEVELOPMENT OF SCIENTIFIC APPROACH 
TO MANAGEMENT 

On first view one might think that recent developments in scientific 
management were independent of work on computers. However, the 
developments are intimately related. 

The concept of "scientific management" is not new. Frederick W. 
Taylor applied scientific methods to certain business operations as 
early as the 1890s. He found particular success in the field of produc- 
tion, since it was there that management attitudes, largely stemming 
from an engineering background, were most receptive. In addition, 
the production process, being largely a physical matter, and more 
directly under company control than certain other activities, proved 
more susceptible to the definition and measurement which are basic 
to any scientific study. 

Taylor hoped to extend his work to all phases of management. But 
despite his efforts, and those of many other workers, this hope was 
not everywhere realized. 

Part of the difficulty lay in the fact that general managers regarded 
their abilities as primarily an art. On the other hand, scientists then 
were not interested in business problems. It was not professionally 
acceptable for scientists to work in commercial fields unless their 
activities were similar to those which would be carried on in university 
research departments. The research departments of companies like 
Bell Telephone, du Pont, or General Electric would be examples. 

World War II was the occasion for government-sponsored research 
leading to the development of computers. The same period saw the 
enlistment of scientists to aid in the solution of logistics and other 
decision problems. A notable example was the use of scientists to 
program British fighter aircraft during the Battle of Britain. Through 
their efforts the limited resources of the British Air Force were utilized 
with unprecedented effectiveness. 

Another such study led to better scheduling of cargo transport ships. 
The mathematical models which were developed for this purpose 
were the direct forerunner of the modern analysis technique known as 
linear programming. 

After the war many of the scientists wished to continue with work 
similar to that with which they had been engaged during hostilities. 



122 Electronic Computers and Management Control 

They had found the problems far more challenging, and therefore more 
interesting, than they had expected. In addition, successful achieve- 
ments during the war led them to hope that similar results might be 
obtained in business. 

Even those who returned to the university life infused a new set of 
concepts into their teaching and thus inspired students to enter fields 
which in earlier years would never have attracted them. 

Since the war, the field of scientific effort in management has con- 
tinued to expand. Two new societies were formed to bring workers 
together, to encourage exchange of ideas, and to further research. 
These societies are The Institute of Management Sciences and the 
Operations Research Society of America. In addition, many other 
business and professional organizations have devoted an increasing 
amount of attention to this field. 

Even so, it has been difficult for many executives to understand 
how the scientific approach differs from that which they have tradition- 
ally used, and why it is of such importance to the future of manage- 
ment, especially as it is related to the new tools provided by electronics. 

Scientific Method. The scientific attitude toward business problems 
can be exemplified by the technical steps which usually are involved. 
These begin with an initial exploration of the problem and an approxi- 
mate formulation of the relationships between relevant factors. This 
is followed by more precise identification and measurement of factors. 
Next, the relationships between these factors and the desired results 
are expressed more formally, often as a mathematical model. Then the 
model is tested by attempting to provide better procedures, processes, 
decisions, and so on. In some cases it can be shown that the solution 
to the problem is an "optimum": no better one can be found under the 
given conditions. In other cases, the model can be improved. 

Not all of management's problems are soluble by scientific methods. 
Nevertheless, some scientists believe that in time much of management 
decision making can be formalized by mathematical models. In this 
sense the use of scientific methods is thought of as leading to a man- 
agement revolution, contrasted with use of electronic computers and 
automation, which are said to be creating a second industrial revolu- 
tion. 

The scientific approach is not confined to the use of any particular 
technique or mathematical method. Nor is use of the approach limited 
to any group of persons such as a team of scientists who are working 



Management and the Scientific Approach 123 

on a particular problem. The scientific method is a general concept 
which anyone can adopt. It emphasizes a logical approach to prob- 
lems with definitions, objective measurements, and formulations, which 
other workers can repeat and test for themselves. 

In practice the use of this approach is often so similar to nonscientific 
methods that it is difficult to make a distinction. In both methods a 
problem usually is broken into manageable parts, these are assessed, 
then the parts reassembled into the solution for the whole. Differences 
lie in the extent to which the problem is rigorously defined, the way 
the measurements of the parts are carried on, and especially in the 
way the relationships between the parts are expressed and the way they 
are formally recombined and repeatedly tested. In business, the proc- 
ess also includes communication of the results back to the decision 
makers so as to provide for action on a continuing basis. 

Installation of electronic computers is not the only force which has 
brought about the need for the scientific approach to management's 
problems. There are many other factors which have encouraged 
executives to seek the aid of scientists. Among these are: 

1. Growth of companies, geographically, in terms of product output, 
number of employees, and so on. 

2. Diversification of product line. 

3. Technological developments, in nuclear physics, electronics, elec- 
trochemistry, etc., including automation. 

4. Changes in the social environment, including changed attitudes 
among employees and executives. 

5. Increase in size of investment required to build new plant and 
equipment, or to undertake new research projects, or to develop 
customer acceptance for new products. 

6. Developments in psychology, "econometrics," sociometrics, etc. 

As a result of these developments there is an increasing need for 
methods whereby managers can rationally select the best course of 
action— one that minimizes risks while taking full advantage of the 
opportunities afforded by changing conditions. New methods are 
required for defining concepts, establishing and analyzing the logical 
relationships between the key factors, and reporting the results in a 
way that will enable the executives to select the best program from 
among the many alternatives available. In addition, a system is re- 
quired that will test the operation in order to determine whether it is 



124 Electronic Computers and Management Control 

proceeding according to program, and that will provide data with 
which to test the program itself. 

Groups which study the use of computers for processing data usually 
become involved with these other problems. By thinking of the 
electronic systems as communication channels, personnel closely in- 
volved with the installation quickly come to realize that attempts to 
solve basic problems in turn have a decided effect upon decisions as to 
the requirements of the data systems and, consequently, on the com- 
puters themselves. 

For example, as one company decentralized its sales operation 
geographically, it found that it required quicker communication of data 
to the home office in order to schedule the central production facility. 
This meant that data had to be transmitted over long distances either 
by messenger, mail, or teletype. The company has found that each 
of the media they can use imposes different requirements upon the 
computer. 

If decision making is decentralized and production is not an urgent 
problem, then the requirements for communicating information to the 
central or home office may be reduced. The only data transmitted may 
be periodic financial statements sent through the mail. For example, 
in one case where completely integrated divisions were established, 
each division required a great deal of data for its own operation, but 
the home office wanted only reports which showed financial operations 
for a period to compare them with the financial budget submitted 
earlier. The company contemplates installation of smaller computers 
at each divisional headquarters. 

Computers may take over a great deal of current scheduling and 
production control as further strides are made in automation. These 
computers may not be entirely similar to the computers described 
earlier for data processing. Some may have analog features (to be 
described later). At the same time, use of such equipment should 
eliminate many of the current requirements for data processing. 

A number of recent applications of scientific methods to manage- 
ment problems have taken place outside the computer field, and 
duplication of effort occasionally has taken place. For example, one 
company recently laid plans to start an operations research group and 
a separate computer-applications group. At the same time, it was 
expanding its analysis and control group for management controls in 
the controller's office. Conflicts were inevitable and have arisen. 



Management and the Scientific Approach 125 

Only a few companies have had sufficient experience to develop a 
well-considered policy in this field. In one case, the computer group 
was transferred from an independent status and placed under the 
controller mainly because as it expanded it entered the operations 
research area. Meanwhile, the operations research department was 
expanded. Management authorized the operations research depart- 
ment to establish several separate research groups and to expand their 
function to include all analysis— long-range and short-range— but to 
coordinate their efforts with the computer group. 

The problems of handling such conflicts are described in Chapters 
7 and 8. The remainder of this chapter will be devoted to an examina- 
tion of the methods of the scientific approach and an appraisal of its 
future. 

Use of Scientific Methods in Business. Why have scientific methods 
received so much publicity of late? What do they really have to offer? 

The businessman whose experience covers a period of time has rea- 
son to be skeptical. Scientific management, as propounded by some 
overenthusiastic followers of Taylor about the turn of the century, by 
the leaders of the "technocracy" boom during the 1930s, and by other 
lesser movements, has promised much without achieving all that was 
expected. Even the punched card was heralded by some as the final 
solution to management's problems. Today the campaign is for man- 
agement planning and control, with "decentralized" management. Yet 
problems persist. 

In what way can the scientific approach improve upon present prac- 
tice in planning and control? What is the management-science ap- 
proach? 

In the first place, there is no new "scientific method," different from 
any which has been employed by some businessmen in the past. The 
difference lies in the fact that the ability now exists to perform in an 
organized and systematic fashion some of the management planning 
and control functions which up to now have been carried on piecemeal. 
Just as new tools have become available for the physical processing of 
data, so now there are new tools for business analysis. These new tools 
are primarily mathematical and statistical in nature. 

No single mathematical method or group of mathematical methods 
exists today which are of primary importance for business use. Tools 
have been used for the solution of particular management problems 
such as: 



126 Electronic Computers and Management Control 

1. Production control. 

2. Inventory control. 

3. Financial management. 

4. Location of distribution outlets. 

5. Transportation choices. 

6. Transportation scheduling. 

7. Investment in new equipment. 

8. Replacement of old equipment. 

9. Executive compensation. 
10. Market forecasts. 

Although certain tools have found most usefulness in one or a few 
particular areas, the capable analyst is always alert to the possibility of 
using any tool which may prove to be suitable. Among his tools are: 

1. Linear programming. 

2. Dynamic programming. 

3. Calculus of variations. 

4. Information theory. 

5. Communication theory, 

6. Sampling theory. 

7. Stochastic models. 

8. Monte Carlo methods. 

9. Servo theory. 

10. Game theory, and others. 

These are not necessarily distinct methods. There is no hard and 
fast line between them. Each represents the work of a group or 
several groups of investigators who usually have concentrated on a 
certain type of problem; history has given a name to their work, 
and thus distinguishes it from other research efforts. As a result, it 
is difficult to classify these tools in a way that is helpful to the business- 
man. Even the terminology used is not consistent. 

But this is not the primary point. Generally speaking, the applica- 
tion of the scientific approach is still in the first stage of experimenta- 
tion and partial application. It may continue in this state for a number 
of years. Much of the work in this field is theoretical. Practical appli- 
cations will follow later. 

This early stage of development offers both an advantage and a 
problem. The advantage to businessmen lies in the fact that refine- 



Management and the Scientific Approach 127 

ment of many of the mathematical tools has not yet progressed to 
the point where it is difficult for the average executive to obtain a 
fair understanding of any particular tool if he is willing to make a 
reasonable degree of effort. The problem arises because there is 
such a variety of tools. It is difficult to foresee which will become of 
significance to him now or in the near future, and which will be less 
significant. 

Knowledge of techniques is less important than understanding the 
purposes that the tools can serve. However, some knowledge of the 
techniques is required in order to achieve this understanding. The 
techniques are predominantly mathematical. They are used to estab- 
lish models. In the present stage of development, the models fre- 
quently depend on oversimplified assumptions. As time passes, models 
will have to be refined. Business will require development of new con- 
cepts, improved methods, new techniques, more suitable units of 
measurements, etc., before a well-developed management science can 
emerge. 

New methods, many of them adapted from other sciences and dis- 
ciplines, are continually being investigated. Since methods are chang- 
ing so fast, study of the field must be continuous, and related to busi- 
ness requirements. 

The evaluation of the newer tools depends upon the ability of the 
executive to develop certain conceptual skills. These skills include the 
ability to see the enterprise as a whole, to know whether objectives 
are mutually compatible and valid, to recognize how various functions 
of the organization depend upon one another, to determine which 
assumptions must be made in order to allocate resources in an optimal 
manner, to recognize various alternative courses of action and their 
potential results, and so on. 

Mathematics is the best tool with which to achieve the coordination 
and communication of such concepts. It can represent conditions 
and events much more concisely and clearly than can be done in any 
other way. 

For this reason mathematics often is called a language. As with a 
foreign language, mathematics sometimes is difficult to learn and to 
apply. But the essence of most mathematical concepts, of a type useful 
to business, is only simple logic. It is not too difficult to learn certain 
principles of mathematics which will enable the businessman to im- 
prove his conceptual skills for managing an operation. Moreover, as 



128 Electronic Computers and Management Control 

the mathematical tools themselves are improved, it is likely that such 

knowledge will become even more useful and perhaps competitively 

imperative. 

Already a number of applications have been made which illustrate 
the use of the tools. A railroad has developed a mathematical model 
of its freight-car operations which was used by the managers to make 
decisions so as to increase the average use of a freight car from 1% 
to almost 3 hours a day. A life-insurance company has made a statisti- 
cal study of the causes of turnover among its younger female employees, 
which enabled the personnel department to cut the turnover by about 
70 per cent. A large manufacturer has pictured its capacity require- 
ments for a new plant in a way that saved 20 per cent in making a 20- 
million-dollar installation. A distributor has rated its profitable items 
so as to increase sales, while using fewer salesmen, and to raise profits 
over 50 per cent. A metals manufacturer has doubled the speed of its 
customer service. All of these and many similar achievements have 
been reported as resulting from applications of the tools of manage- 
ment science. 

In a number of cases the tools were used to describe situations which 
the businessman already understood fairly well. In such cases trans- 
lation of the manager's concepts and ideas has served primarily as a 
means for experimental test of the models. But more often the situa- 
tions were not so well understood as was thought at first. Use of 
mathematics for these problems has established the concepts more 
clearly, and sometimes has uncovered concepts which were embedded 
in long-standing practices or rule-of- thumb judgments. Thus, using 
scientific methods, the executive has been able to refine his concepts. 

An example is the application of mathematics to production and in- 
ventory control so as to add exactness to the concept of "exploding" 
a bill of materials. By stating requirements mathematically, improved 
systems have been developed with which to pass information from one 
department to another. These systems short-cut former procedures, 
which actually were more complicated than the work of developing 
the mathematical models and using them. The old systems could be 
made to work only through much human effort and expediting. 

Improvements in Method. Scientific attempts to improve the quality 
of management decisions through use of more advanced mathematical 
techniques are now under way in a number of companies. The in- 
creased use of statistical sampling is an example. When a mass of 



Management and the Scientific Approach 129 

accounting data is involved, satisfactory managerial conclusions usually 
can be drawn from studies of much less than 100 per cent of the data. 

Other examples of modern mathematical developments are easily 
found. Some mathematicians are attempting to record and analyze 
data concerning events through the use of matrix algebra. New con- 
cepts of quality control, based on statistical measurements, are being 
applied to the industrial process. Statistical methods, known as design 
of experiments, are being used for such matters as study of the inter- 
relationship of the production processes and the quality of the product. 

In the past, most methods of work measurement and production 
analysis have attempted to measure performance at many individual 
stages and to use each measurement to control the process at that 
stage only. The new approach is to attempt to find out more about 
the whole production process, and how changes in particular manu- 
facturing methods can affect it. This is done through the use of 
small samples scientifically selected and analyzed as a group. More 
precise answers can be obtained in this way at less cost. If the system 
is properly designed, the analyst often can call management's attention 
to the need for taking action while the process is still in operation and 
before it goes out of control. 

In the near future it is expected that more and more monitoring for 
this purpose will be done by automatic measuring devices linked to 
electromechanical control systems. But this development will be pos- 
sible only when the relevant production factors can be identified, 
measured, and the relationships between them clearly expressed. 

Scientifically formulated concepts cannot be developed for the solu- 
tion of all types of business problems. Where the factors for decision 
are not clearly identifiable and measurable, where useful relationships 
between them cannot be established, then human judgment will have 
to be used, as it customarily is at present. These situations, many of 
which are primarily concerned with personal relationships, constitute 
the bulk of business decision making today. Nevertheless, the scientific 
approach can provide a better analysis of those factors which can be 
identified and measured, and it can lead to a more objective analysis 
of the remainder of the situation. 

Application of Scientific Method— Case Study. The recent experi- 
ence of one company illustrates a number of important points which 
are involved in the application of scientific methods to business prob- 
lems. One of the first matters of concern in the scientific approach is 



130 Electronic Computers and Management Control 

establishment of the frame of reference for the situation. This deter- 
mines the direction in which the investigation should proceed. In 
the company cited the major problem was improvement of the quality 
of the product, without substantial increase in cost. The analysts 
concentrated first on the technological problems, and then turned to 
managerial aspects. 

The first step called for a thorough understanding of the character- 
istics of the product, the specifications which customers wanted, and 
the way in which the production process attempted to meet these 
requirements. The purpose of each production activity had to be 
determined as it related to achievement of the desired characteristics. 
The required tolerances at each point on the production line had to 
be determined, units of measurement for this purpose had to be 
chosen, and the best methods of production ascertained. 

Much of the project time was spent on this topic. The analysts 
investigated actual shop conditions, studied reports, and discussed with 
research and production personnel the reasons for the behavior of the 
various physical processes. From this investigation the group set up 
a conceptual model of the way to make a high-quality product. The 
conceptual model was first established in rather general, nonmathe- 
matical terms, and its validity tested both by discussion and by using 
it for suggesting changes in the production process. 

Following this, the model was reduced to mathematical expression, 
which made it possible to vary different factors and to measure their 
effect on the characteristics of the final product. After careful analysis 
it was possible to make definite statements as to what improvements 
should be made in the production process, which factors were critical, 
and the degree of control which should be exercised over them. 

The second step involved consideration of market demand and other 
managerial requirements, such as minimization of inventory levels. 
The model was expanded to take these factors into account. Now 
the analysts could determine how the product should be built, in view 
of the quantities of the various classes of product desired. Using 
the model, it also was possible to determine how production tolerances 
at specific points should be altered in accordance with changes in 
demand for various classes of items. 

The significance of this type of study does not lie in the use of any 
particular mathematical or statistical technique. In research work, 
often there is a danger that a technique may be applied for its own 



Management and the Scientific Approach 131 

sake. Someone learns that a particular technique has produced a 
helpful result in one situation, and so he attempts to apply it to an- 
other, without sufficient study of its fitness. Ability to define the 
objective is fully as important as the ability to apply particular tech- 
niques. 

A major obstacle to adoption of the scientific approach seems to be 
a gap between management, who have the ability to define problems 
and to rate their relative importance, and the scientific analysts, who 
know the tools but may try to apply them without fully understanding 
the significance of the problem or without being able to communicate 
with the managers in order to find this out. 

Several other aspects of the situation merit attention. To complete 
such a project successfully it is necessary to have top-management sup- 
port, acceptance by operating management and personnel, and a 
capable team that has the proper balance between experience and 
training, and whose members can work together. 

The team must be able to take advantage of a scientific approach. 
Partial answers are not sufficient. Apparent solution of a part of the 
problem may not improve the over-all picture. For example, if the 
quality of the raw material (in the example discussed) had been 
further improved, the cost of raising the quality would still have been 
more than the increased revenue. Major improvements, which actually 
increased the profits, were only to be found in the production process. 

In addition, the study did not stop with the physical flows and 
activities but went on to consider the profit-planning aspects of the 
situation. For this it was necessary to establish clearly and precisely 
the data needs for management planning and control at all levels of 
responsibility. At this point the requirements for electronic data 
processing finally could be specified. A model, showing expected data 
flows, could be built. 

Model Building. Earlier in the chapter it was stated that the scien- 
tific method is characterized by careful description of the entities in- 
volved, measurement of the relevant factors, and establishment of the 
relationships between them in a conceptual framework which can be 
tested and improved by experiment or observation. Perhaps the best 
way for a businessman to visualize this process is in terms of "model 
building." 

A model is a representation of reality. It tries to retain as many of 
the important characteristics of the situation as can be handled by the 



132 Electronic Computers and Management Control 

investigator. But it leaves something out in the interests of simplicity. 

A physical model of an aircraft may look so much like a plane that 
a person who has seen only the model may recognize the actual plane 
the first time that he sees it. If this recognition is all that is required— 
as for recognition of enemy planes in combat— then the fact that the 
model left out many of the internal features of the real plane is im- 
material. 

Similarly, a mathematical model or a work-flow diagram is a sym- 
bolic representation of the structure of the business problem. The 
model attempts to state all or most of the relevant structural, quantita- 
tive relationships. But inevitably it leaves out some of the details, 
again in the interests of simplicity. Relative simplicity is required for 
effective communication and analysis. 

What is the value of such models to a businessman? In what way 
does mathematical representation of relationships hold promise for 
making better decisions? 

Most business decisions are made with the assumption that many 
past relationships will continue. It is expected that what has hap- 
pened before will continue to happen, barring unexpected influences 
on the situation. Departmental cost comparisons now drawn up in 
business, for example, are based upon this type of assumption. The fig- 
ures for this year and for last year, same month, or for this month and 
for last month, are listed in adjacent columns on a report. Obviously 
the comparison is meaningless unless the assumption can be made that 
the two periods have some continuing relationships, despite obvious 
changes. Any explicit formulation of these relationships is a model. 

Construction of a model of the work flow is the first step in studying 
a business routine for application on a computer. The various indi- 
vidual operations of a program usually are quite numerous. The rela- 
tionships between them are complex. It is not possible for the pro- 
grammer, however skilled, to keep all of the details in mind. The 
problem must be sketched out in terms of the major flows. These are 
the input sources, tape units involved, major processing steps, the types 
and quantity of storage involved, and the output. Each of these then 
can be approached as a separate problem. The programmer must keep 
in mind the necessity for integration of each with the other major sec- 
tions of the work-flow model, but he can concentrate on details in a 
narrower area. Each step thus drawn up is a submodel. In this way 
a "picture" of the process is laid out for study and analysis. 



Management and the Scientific Approach 133 

There are many types of models and many methods used to develop 
them. Because of this diversity, it is useful to classify models into three 
general categories— descriptive, predictive, and decision. 

Descriptive Models. A descriptive model states in condensed, 
mathematical form the measurable relationships which appear to exist 
in a business situation. It attempts to portray the situation as it exists, 
either during a period or at a point in time. As observations indicate 
relevant differences between the model and the actual circumstances, 
the model is changed so as to provide a more accurate description. 

The best known and most useful of descriptive models in business is 
the accounting system. The accounts are so closely connected with 
business procedures that it is sometimes forgotten, especially by ac- 
countants, that the accounting system is only a symbolic representation 
and not the actual circumstance. It is important to remember that the 
purpose of accounting is to mirror, or "model," the business situation 
so that managers will be able to understand and to control it. If the 
two are not in line, it is the accounting system and, specifically, the 
amounts in the accounts, which should be adjusted. 

The cost system is a separate model. Although usually tied in with 
the regular accounting system, actually it can be held quite separate 
from it. Cost accounting starts with a group of cost figures, classified 
according to form ( materials, labor, taxes, rents, depreciation, supplies, 
including all the "overhead items"), and, based upon some analogy 
with an observed physical relationship of these resources to the prod- 
ucts or departments in which they are used, reallocates the costs into 
product or department groupings. 

Other descriptive models are frequently used in business operations. 
Fundamentally the systems of production control, inventory control, 
time study, the inspection and testing procedures, sales analysis, and 
so on, are descriptive models. 

A descriptive model which has become of increasing importance to 
businessmen in recent years is that which has been developed to repre- 
sent our national economy. Various national-income concepts, and 
especially such items as "Gross National Product" and "Personal Dis- 
posable Income," etc., have been found useful in assessing the current 
economic situation and in attempting to forecast how the course of 
business will affect the sales and other activities of a particular com- 
pany. 

In Chapter 8 and in Appendix 4 a more detailed discussion will be 



134 Electronic Computers and Management Control 

devoted to a new method for constructing a model of an entire re- 
porting system. 

Predictive Models. Any descriptive model can be made into a 
"predictive" model if it incorporates a forecast of one or more of the 
factors. If the model is sufficiently accurate, the other factor or factors 
then can be predicted. In other words, since the model shows the 
relationship of the other factors to the forecast factors, an estimate of 
the latter in effect also makes an estimate of the former. 

Predictive models still do not indicate the course of action which 
should be taken. The manager must make this choice, using his judg- 
ment. However, since most management problems deal with future 
events, to the extent that important considerations can be predicted 
the models are extremely useful. 

The most common of predictive models in business are budgets. 

A budget is a formal plan for the future operations of a business. 
The purpose of the plan is to establish some sort of prediction for each 
particular operation, and to coordinate the various individual opera- 
tions so that each responsible person will have an idea of what is ex- 
pected of him. The total budget is then an expression of the total 
expected expenses and revenue, with the resulting profit or loss— or, in 
the case of a cash budget, the expected receipts and disbursements. 

Budgets can be prepared because past experience gives those in 
charge a pattern of relationships with which to work. If certain items 
are estimated or arbitrarily set, then others can be constrained or will 
tend naturally to fall within certain limits. For example, if sales are 
to reach a certain amount, then a certain amount of goods must be 
available, which means inventory must be produced or purchased. A 
certain number of salesmen must be hired and paid, a certain amount 
of advertising must be ordered, and so on. The relationships are not 
exact, but unless policies and practices are drastically changed, they 
will not vary substantially within short periods, for most businesses. 

In mathematical terminology, budgets are built up from the "inde- 
pendent variables" in the situation. The other factors are "dependent 
variables." The former are the basic estimating factors, the forecasts 
of the strategic elements in the business situation. The latter are ex- 
pected to follow if the forecast is correct. 

Predictive models of the logistics problem of the Department of 
Defense— its supply schedules, etc.— have been formulated by a number 



Management and the Scientific Approach 135 

of groups in the various armed services. Partial models have been 
prepared by researchers at universities and for several private com- 
panies. For example, publications of the RAND Corporation, the 
Cowles Commission, the Management Science Project at UCLA, and 
the Graduate School of Industrial Administration at Carnegie Tech are 
frequently concerned with this type of model building. The Air Force 
and the Navy are engaged in testing and applying the findings of these 
groups. 

Much of the experience gained in the military will be directly appli- 
cable to industrial problems. Among companies which have shown an 
active interest in production schedule models are U.S. Steel, SKF, 
Lockheed, and Kaiser. 

An example of a predictive model is the study which was made for 
the Seebury Farms, which grow vegetables for the Eastern market. 
An analysis of climatic conditions, market demand, growing rates of 
vegetables, availability of help, and other factors resulted in a specific 
planting program. This was worked out so that a rotation of crop 
maturities was integrated with the other factors. As a result, overtime 
was reduced, unsold inventory minimized, product quality increased, 
and the general profit picture improved. 

The basic concepts are applicable to the problems of other farmers, 
but unless all their conditions are similar, a new study of the factors 
relevant for their purposes would have to be undertaken. Of course, 
experience gained in the Seebury situation would be of help. 

Decision Models. A group of predictive models, together with some 
means of choosing from among them the one which fulfills some objec- 
tive most completely, is called a "decision model." 

The primary function of decision models is to tell the user what to 
do in view of some stated purpose. For example, the manager may 
desire that future profits be maximized. His decisions are subject to 
certain measurable limitations, such as the storage facilities available, 
transportation costs, transportation times, and so on. Given these con- 
ditions, there is a set of decisions which will yield the best profit. 

Models which determine these decisions have been used to tell the 
businessman where to ship; what, where, and how to produce; and 
where to sell. These models often use the new technique known as 
"linear programming." 

Decision models can also be constructed to determine such matters 



136 Electronic Computers and Management Control 

as the best pricing policy or the best financial structure. They show 
the effect of various long-range programs on the resources of the com- 
pany, on the finances, on the production facilities, and on the man- 
power. Given some "optimizer"— a method of determining the most 
desired procedure— the models can be used to choose the resource allo- 
cation which will produce the most profit. 

The range of problems susceptible to decision model building has 
not been fully explored as yet. Much work has been done, but most of 
this work is on individual types of problems, primarily those which 
occur at the operating level. Even here problems such as those which 
involve scheduling on individual machines have not lent themselves 
too well to model building. It is difficult to establish a model where 
work flow must be scheduled before or after a specific machine tool. 

Many researchers are of the opinion that, even if such models could 
be built, they might be of limited value because of the way they would 
limit the foremen's initiative. The models would have to be extremely 
complex, so that they could take the various pertinent factors into 
account. Examples of these factors are machine breakdown, tool break- 
down, changes in material, sickness of the labor force, strikes, etc. 

The brightest future of decision models lies in the area of manage- 
ment planning and control. Management planning and control is pri- 
marily concerned with exploring future conditions. But executives are 
hindered from examining alternative plans for the future, because of 
the limitations of present data-processing systems. Applications of the 
new concepts of information theory and game theory have been handi- 
capped by the primitive state of our knowledge of business decisions- 
how and why they are made, etc. Usually it is all that a company can 
do to prepare one financial budget within a given period. To prepare 
a decision model, based on a number of budgets, will require the use 
of electronic systems of communication, electronic computers, and new 
mathematical methods of analysis. 

The Future of the Scientific Approach in Business. A model has 
no intrinsic value. It should be constructed only if it is useful. On 
the other hand, the fact that the model may not be as elegant as one 
might wish does not destroy the value of the model if it is properly used. 
A number of critics have assailed the use of models in business, since 
it has been found difficult to measure and particularly difficult to fore- 
cast items as precisely as these critics desire. 



Management and the Scientific Approach 137 

The critics point to the greater precision which has been attained in 
the physical sciences and declare that, until such precision is reached 
in the social sciences, researchers have no right to call their techniques 
scientific. 

Admittedly the degree to which identifiable factors can be measured 
—compared to the influence of unidentifiable factors— does help deter- 
mine whether or not an approach can be scientific. It acts as a limit 
on the area where scientific methods can be applied. Precision in 
model building is related to the difficulty of the problem and the state 
of human knowledge concerning specific techniques and their appli- 
cation. 

Business problems have proved to be much more complicated than 
most scientists ever realized. For example, during the war a group of 
scientists developed a rather remarkably effective "search theory" for 
the location of submarines. After the war some of them assumed that 
the theory would be applicable to supposedly simpler problems of cus- 
tomer location for retail operations. It wasn't. 

But the fact that usefulness of the methods is limited at present does 
not mean that they can be safely ignored. The value of the scientific 
approach, with the development of models, can be judged on two 
grounds: 

1. Is the scientific approach more effective than present practice? 

2. Do the models provide a conceptual framework from which con- 
tinued improvements can be expected, as the dynamic cycle of 
"concept-observation-improved concept-further observation" pro- 
ceeds? 

In answer to the first question, it is already apparent that useful 
results have been achieved. The publications and conventions of pro- 
fessional societies have featured many reports of successful applications 
of scientific methods to business. 

A survey of over three hundred businessmen, reported by Caminer 
and Andlinger in the Harvard Business Review, showed that the appli- 
cation of scientific methods to business management is a matter of 
major interest to executives. The executives' greatest complaint was 
the lack of case histories in their particular fields of interest. As the 
authors pointed out, a reason for this lack is the fact that many com- 
panies with experience are reluctant to release this sort of information. 



138 Electronic Computers and Management Control 

The position is understandable, since the studies usually are concerned 
with critical policy decisions. Information concerning these would be 
of assistance to competitors. 

As for the second question, the trend toward automation of both 
office and factory will make business processes more quantifiable and 
therefore more subject to analysis and accurate prediction. At the 
same time, some means must be found for integrating the massive flow 
of information with which the modern executive is confronted. Scien- 
tific methods of analysis seem to offer the best solution of this problem. 

In the same survey, executives also expressed opinions on this point. 
They saw a growing trend toward use of this type of research in all 
areas of American industry. 

The essence of the scientific approach in business is the develop- 
ment of policies and procedures that can be continually improved 
through observation and experiment. 



CHAPTER SEVEN 



Management Planning and Control 



Electronics and mathematics already have been used to improve a 
number of business routines and to solve certain specific problems. 
But though such use has saved money, partial applications are not the 
most profitable use of these new tools, nor are they the goal of the 
businessmen who are developing their use. 

Several different terms have been employed to signify the anticipated 
"best use" of electronics and mathematics. Among these terms are 
"integration," "the systems concept," "profit maximization through com- 
mon-language systems," "clerical as well as factory automation," and 
so on. The terms are used to describe efforts that are being made to 
apply the new tools to the management of the whole business, not 
merely to particular parts of it. 

The developments are taking place in management planning and 
control. They have provided managers with a series of new concepts. 
The concepts enable managers to formulate company goals and require- 
ments in a more definite pattern than has formerly been possible. To 
do this it is necessary to integrate company objectives, organization, 
the method of information analysis, the system of communication, and 
the whole decision process. 

Significance of Planning and Control. In recent years the term 
"planning and control" has received considerable attention from man- 
agement. There are two aspects to this term: its general meaning, and 
the set of concepts which have come to be associated with a particular 
use. 

The more general meaning of "planning" covers the process whereby 
top management sets an objective, organizes the activities, establishes 
the authority-responsibility relationships of the managers, staffs the 
organization, and commences the direction of operations. 

139 



140 Electronic Computers and Management Control 

"Control" compels events to conform with plans. It is the process of 
setting specific standards in relation to specific allotments of resources, 
followed by measurement of progress, comparison with the standards, 
and making adjustments as needed in order to achieve the original 
objective. Reported results are then used in future planning. 

In its more particular meaning the term "planning and control" has 
come to be associated with a revival of interest in the profit motive. 
A number of writers, such as Keith Powlison, in the Harvard Business 
Review, and Perrin Stryker, in Fortune, have called attention to the fact 
that a company's top executives are likely to put several other objec- 
tives ahead of making profits. Among these are the desire to be the 
biggest, or most secure, or best ( in a number of ways, down to items 
such as having shops that cut metal with the least amount of scrap, 
regardless of efficiency). Some executives would put "statesmanship" 
first— the desire to improve social conditions directly, rather than rely- 
ing on competition, or on supply and demand. 

No one will argue that businessmen should turn away from the need 
for giving consideration to such factors as the social effect of their 
decisions, or that they should look only to the profit figure. The ques- 
tion is rather one of weighting a number of objectives. How much can 
businessmen afford to pay for the attainment of these other objectives? 
How should managers relate them to the profit picture? 

This aspect of business decision making has become more important 
in recent years, for a number of reasons. Rapid corporate growth has 
made management realize that their companies were becoming too big 
to administer without better systems of planning and control. All 
companies, regardless of size, have encountered drains of profit dollars 
as the result of higher taxes, labor demands, and the need for expansion 
and replacement of facilities. Technological developments and prod- 
uct diversity have also shown the need for greater emphasis on profit- 
able decision making. 

While reemphasis on profits by itself often leads to an increased 
interest in planning and control, the pressure will become even greater 
if businessmen try to take into account the full range of their objectives 
when making decisions. Planning and control means more than tech- 
niques and tools. It is a concept of good business management. 

Dr. T. F. Rradshaw, an authority on the subject, has outlined the 
profit approach: 



Management Planning and Control 141 

By profit planning and control I mean the acceptance of certain objectives. 
These are the objectives: first, setting a profit goal; second, setting depart- 
ment goals which, taken together, will achieve the profit goal; third, 
measuring progress against these standards; fourth, making continual ad- 
justments to keep the whole organization moving in balance toward the 
goal. 

To accomplish the objectives of this predetermined goal in the fullest 
sense, it is necessary to coordinate not only the full range of objectives 
with the tools of planning and control, but also to integrate them with 
the organizational structure, the employee incentives offered, the pro- 
gram of management development, etc. Apparently few companies 
have understood this problem, and so far it has been too early to apply 
scientific methods for a solution. 

DIFFICULTIES IN ATTAINMENT OF PLANNING AND 
CONTROL OBJECTIVES 

While it is easy to state the concept of profit maximization, to put 
this type of planning and control into practice is a project of major 
proportions. The new system would affect all aspects of the opera- 
tions, all parts of the organization. 

Psychological and Technical Difficulties. There are always psycho- 
logical resistances to be met among the people affected; the communi- 
cation fails at all levels. There are technical difficulties of measure- 
ment, recording, and analysis. 

People resist change in this direction for several reasons. They 
resent the fact that they are to be judged and guided more closely. 
The tools of planning and control are regarded as setting up a system of 
personal score cards, which too often are used as mere pressure devices. 

Planning and control projects have run into a number of technical 
difficulties. In most cases the accounting data are used as a start. For 
managerial purposes, these data may not be too relevant. More often 
than not, the accounting process does not produce the data in time, or 
in a manner suitable for use in making operating decisions. 

Operating personnel usually are accustomed to thinking of produc- 
tion in units rather than in dollar figures. As a result, communication 
fails. A frequent result is that operating managers keep a set of little 
"black books" for their own use in making decisions. These "black 



142 Electronic Computers and Management Control 

books" also reflect what they believe will enable them to defend them- 
selves, as well as to please their superiors. 

Conflicting Requirements. Another difficulty in establishing a sys- 
tem of planning and control arises from the fact that different levels of 
authority have different uses for the same data. They place different 
emphasis on them. First-line supervisors frequently use accounting in- 
formation as a mere fill-in on those aspects of the work that they 
cannot appraise from their actual contact with it. The more removed 
they are from the scenes of operations, however, the more importance 
they place on accounting data as a means to judge subordinates and 
to check on what is happening at the operating level. A planning and 
control system which is well organized to provide data for the one pur- 
pose usually is not able to provide adequate data for the other. 

If the upper-level executives are to use the reports for planning and 
control, the operating executives must be satisfied that the data are 
accurately recorded, that the expected standard level is reasonably at- 
tainable, and that the variables which the system measures are con- 
trollable by them, at least in part. Otherwise, a great deal of time and 
energy will be used up in debate about the accuracy of the accounting 
charges or about the circumstances that explain the variances. 

To obtain this confidence, a close and direct relationship between the 
operator and the accountant is necessary. This must be reflected in 
the organization structure. 

For many problems it is difficult to forecast the types of data which 
will be useful. There are two ways that this difficulty may be met, two 
ways to increase the usefulness of the data. The first is to elaborate 
the regular reports, so that they include more details. The other is to 
have a special staff whose duty is to investigate and gather the facts 
that are needed for a specific problem. The latter method often has 
been found to be the more promising. 

Both methods run into the problem of timeliness. Detailed regular 
reports require the processing of vast amounts of data, and some staff 
studies still are required to spotlight needs and to sift the information 
available for the pertinent facts. 

On the other hand, special studies take time, and the range of prob- 
lems covered at any particular time is necessarily limited. 

Inadequate Means for Evaluating Performance. It is difficult for 
management to assess the benefits of a particular system of planning 



Management Planning and Control 143 

and control. Most of the yardsticks available are part of the system 
itself. The only over-all measurements managers have are the trend of 
the profits through time, and comparison of their profits with those of 
other companies. But it is difficult to tell how much of an improvement 
is due to the system of planning and control, since so many other cir- 
cumstances also change. Comparison with other companies is some- 
times helpful, but again circumstances are usually so different that 
conclusions can only be tentative. 

The "profit" yardstick implied in "planning and control" is not a 
short-run affair. Obviously no company can hope to prosper if the 
country as a whole, or in many cases the company's particular locality, 
is falling into severe depression. Increases in the standard of living for 
the population is an almost necessary concomitant of the growth that 
most companies strive for. In this sense it can be said that contribu- 
tions to social welfare do have a long-range "profit" value to a particu- 
lar company. 

Obviously, however, such contributions are very difficult to measure, 
and in present practice little attempt can be made to evaluate them. 
That companies recognize this responsibility, however, can be seen in 
such things as charitable grants and support of education. Occasion- 
ally directors have been taken to court by stockholders' suits seeking 
to prevent such activities on the shortsighted ground that the gifts 
cannot be shown to be of benefit to the company. However, it is estab- 
lished that fairly sizable amounts for such purposes can be justified, for 
their public-relations value as well as for the larger objective of social 
welfare. 

On an inside-the-company level the same difficulty of measurement 
has appeared. It is theoretically certain that such matters as employee 
morale, the organization structure, personnel relationships, the plan- 
ning and control operations themselves, and so on, all contribute to 
long-range profits. But how this can be translated into standards for 
the evaluation of individual managers' performance is a problem which 
as yet has not been solved. 

The higher the manager's position, the more important his ability 
to control these intangible factors becomes. Yet top-management 
standards are still a largely unexplored area. Some work has been 
done by groups such as the American Management Association and by 
business management-consulting firms. But the field has received virtu- 



144 Electronic Computers and Management Control 

ally no attention compared with the concentrated effort made by groups 
such as those in production management with their time and motion 
studies for factory operations. 

Most of the standards developed so far relate to physical quantities, 
cost standards, or financial ratios. Physical standards of output, man- 
hour utilizations, and so on, are well developed, but companies still 
can benefit from further investigation of this area. Many can benefit 
by learning to ignore the dollar figures supplied by accountants, when 
prices are not under the control of the operator. This alone would 
simplify and speed many reports. It would recognize the fact that as 
yet most companies are not able to "profit plan" on the operating-fore- 
man level. 

Cost standards have received considerable attention in recent years, 
and there is a decided trend toward relating "cost centers" to authority- 
responsibility units. However, the relationship of these units to the 
over-all company operations, in terms of their profit contribution, is a 
problem that has been avoided— or, as in the case of one large printing 
establishment, "solved" by adding arbitrary percentages of "profit" 
to the work-in-process costs as they move from department to depart- 
ment. 

Financial standards involve the use of ratios, of turnovers, of asset- 
liability relationships, of capital commitments, and so on. These are 
virtually the only quantitative controls used by top managers in many 
companies. But changes in turnover, or in asset-liability relationships, 
are seldom specifically related to profits. 

Capital commitments frequently are justified in terms of cost savings 
rather than profit expectations ( except for very large commitments in 
new plants or new products ) . In fact, it is only in an instance such as 
the study of the effects of introduction of a new product that the com- 
pany-wide profit effect is likely to be presented, together with careful 
quantitative analysis to support the recommendations. In later evalua- 
tions, however, the "profit" contribution of the new product again is 
made on a cost-accounting basis. Cost accounting measures the direct 
revenues and expenses, plus an allocation of other expenses, without 
demonstrating the marginal or other effects such as those resulting from 
the decrease in attention paid to other products. 

The top-level profit standard which has received most attention in 
recent years is the "return on investment" standard used by du Pont 
and General Motors. The success of these companies has led to con- 



Management Planning and Control 145 

siderable publicity for this type of analysis, and it probably represents 
the highest level of profit-planning analysis which can be achieved 
using present systems. The approach is to assume that the over-all 
company profit will be satisfactory if each decentralized operating di- 
vision can earn a satisfactory return on the capital entrusted to it. 
Then, in the absence of negative evidence such as strikes, price-level 
changes, etc., the manager is assumed to have been successful. 

Organizational Problems. The difficulties inherent in a profit-ori- 
ented system of planning and control are highlighted by the problem 
which many companies are facing at present: What is the best organi- 
zation to achieve their profit objective? These companies have found 
that they cannot progress, or in some cases even survive, in today's 
competitive economy if one or a few men try to do all the planning and 
control. 

Companies such as H. J. Heinz, Ford Motor, and Chrysler grew 
and prospered under the personal management of a strong individual. 
However, these and other similar companies encountered difficulties. 
The first of these came with the death or retirement of the dominant 
individual. In few cases had adequate successors been trained to 
continue the centralized direction of the enterprises. Even where this 
had been done, however, it frequently was found that conditions in 
the industry had changed. No longer could one man adequately plan 
and control the diverse operations. Technological specialization, as 
well as advances in the techniques of management, had come into the 
picture to the extent that no one person could be expert in each field 
and relate all of them in the decision-making process. 

In the typical large company, staff experts in the various fields of 
management, such as personnel, systems work, advertising, quality con- 
trol, industrial engineering, economic forecasting, and so on, have 
found that to operate effectively they have to "have the ear of the 
top man." Since the experts in purchasing, production, sales, research 
and development, and so on, are also reaching for the attention of 
the top, the executive has to pick and choose. He must select from 
the problems that are brought to his attention. But he cannot listen 
to them all. Moreover, some problems are deliberately hidden from 
him. As for the problems he does work on, sometimes he cannot get 
all the information needed in order to make his decision; sometimes 
he is overwhelmed by the mass of materials offered for his considera- 
tion. 



146 Electronic Computers and Management Control 

The pressure of day-to-day problems on the top executive is so 
great that too frequently he has insufficient time to devote to long- 
range planning. Such planning, therefore, takes a back seat. It is 
often delegated by default to subordinates in a staff position. These 
men may not understand the operating man's problems, nor do they 
always fully comprehend the significance of other factors, such as the 
company's research programs, public relations, and so on. 

Heinz, Ford, Chrysler, and many others have sought a solution to 
this problem in decentralization. Such companies usually start the 
process of giving more authority to managers by divisionalizing their 
activities. They bring in new personnel, particularly for planning and 
control. This type of program usually has shown results. 

But the basic problem, optimum planning for profit maximization, 
has not been solved, and new problems have been created. In other 
words, the problems have been placed down a level so that now each 
divisional manager must try to solve his problems and also do his 
share of the planning while trying to take into account all the factors 
that top management supposedly should consider. 

As a result of the lack of complete planning at the top, plus failure 
to communicate effectively those plans which are made, the divisional 
manager seldom feels certain that he knows exactly what is expected 
of him. Even when he is given a "rate-of-return" goal, as in the well- 
known du Pont-General Motors system, he knows that in actuality this 
measurement will be weighted— as it should be— by other considera- 
tions, such as community relations, development projects, personnel 
training, competition in his particular field, and so on. He never can 
be sure just how much these other considerations will count. 

Decentralization has spotlighted these problems. It has awakened 
many managers to the need for a reexamination of their planning and 
control concepts. Peter F. Drucker, in a speech before the National 
Society for Business Budgeting, has stated: 

Then take the trend toward decentralization of industry. This is a 
very misleading term because it obscures the fact that you cannot de- 
centralize without a very much better controls of performance, very much 
better controls of objectives, than most businesses have ever heard of. 
Decentralization means that you set up a unit autonomously in such a 
manner that it operates against standards, objectives and goals rather than 
against supervision. The major problem is a control problem which very 
few businesses either understand or have the controls for. Many who go 



Management Planning and Control 147 

ahead and decentralize are going to rue the day. Yet, I believe it is a 
desirable trend. 

Summary. Proper organization for planning and control should 
emphasize neither centralization nor decentralization. The long-range 
plans and objectives of a company, with their related controls, must 
be integrated. The operating decisions concerning short-run profit 
objectives, with their related controls, should be decentralized, but 
related to the central plans. 

Regardless of the problems and difficulties inherent in planning and 
control, it has a great potential for American business. Planning and 
control does aim to let the management know where their companies 
are heading, profitwise. It does attempt to provide an integrated view 
of the business. 

Modern techniques of communication— using electronics, and of data 
analysis for decision purposes— using a more scientific approach, are 
directed toward an improvement of planning and control. A discus- 
sion of this development is presented in the next chapter. 



CHAPTER EIGHT 



Programming, Scheduling, and Feedback 



In order to indicate clearly the essence of the suggested future de- 
velopment of planning and control, several new terms will be intro- 
duced. The same general area is involved, but the approach to busi- 
ness problems will be sufficiently different so that confusion may be 
avoided by the use of new terminology. The terms proposed are 
"programming," "scheduling," and "feedback." 

The selection of the best plan or budget for a given company, one 
which recognizes alternative combinations of the basic planning fac- 
tors, will be called "programming." * Determination of the timing, the 
specific methods to be employed, and the specific resources to be used 
in carrying out the program in a coordinated and consistent manner 
will be called "scheduling." Communication of current information 
to the proper organizational level for analysis, action taking, and de- 
cision making with regard to both programming and scheduling will 
be called "feedback." 

PROGRAMMING 

Programming is a more scientific approach to planning. It is a 
formal, objective attempt to define, measure, and analyze all the rele- 
vant factors in the business situation, and to develop thereby the 
specific policies and plans which will best lead the company to attain- 
ment of its objectives. Programming includes simultaneous considera- 
tion of factors from the various administrative functions, such as plan- 
ning, organizing, staffing and direction, and, to some extent, controlling, 
as well as from the operating functions of selling, purchasing, produc- 

* "Programming," as used here, is not to be confused with the use of the same 
word in connection with operation of an electronic computer— the "program" 
of instructions given a computer. 

148 



Programming, Scheduling, and Feedback 149 

ing, etc. It is best exemplified by the mental processes of men like 
Ford, the du Ponts, Eastman, McCormick, and Swift, while they were 
directing the growth of their companies. They were "into everything." 
They learned that they had to have an acquaintance with all the factors 
in the business situation, since different groupings of them became rele- 
vant for the solution of different problems. 

Present need for the programming concept arises from the fact that 
companies have found that they can no longer successfully centralize 
all important decisions in one or a few top executives. Decentraliza- 
tion of some decisions solves this problem only to introduce the need 
for centralized decision making of a different sort. 

Programming attempts to solve both problems. It tries to take into 
account all relevant factors, and to spell out objectives in such a way 
that decentralized managers will be able both to contribute to and 
participate in company planning. Also, they will receive directions 
from the top management, which they in turn can use to establish their 
own plans. 

The programming concept is important for effective decentralization 
of short-range decision making. It is entirely possible for a company 
to have each of its decentralized managers operate his department with 
a satisfactory return on investment, yet have the company as a whole 
not make full use of its resources. 

As an actual instance, a processor found that after granting consider- 
able authority to managers of decentralized districts, price wars on 
their own products would develop in fringe areas between the dis- 
tricts. The company as a whole suffered, though individual managers 
sometimes made gains. 

In another case a large chemical company found that in decentraliz- 
ing it had created the opportunity for a considerable duplication of 
effort. Each district was repeating, often with somewhat different 
objectives, many of the central staff operations of the company and 
even some of the production and sales efforts. The company has an- 
nounced that the heads of the various divisions have been relocated in 
the central home office as the first step in combating this problem. 

Both of these companies are in the forefront of experimentation with 
electronic systems and mathematics. 

Programming Requirements. Programming uses decision models to 
allocate limited resources among a number of competing demands so 
that the plan will result in the optimum use of resources. 



150 Electronic Computers and Management Control 

When there are only a few alternatives, then the right answer often 
can be selected quickly. However, when the number of variables 
becomes larger, as when a number of divisional managers compete for 
the resources, then selection of the right answer is very difficult. The 
problem of optimum selection becomes still more complex and time- 
consuming when forecasts of future events also must be taken into 
account. 

Company Objective. Selection of the best program requires that the 
objectives of a company be set forth in a manner which is quantita- 
tively measurable. Profits are always stated quantitatively, in dollars 
or as a rate of return, but other objectives also will have to be stated 
in quantitative terms so that they can be interrelated. 

For example, the personnel policy of not discharging faithful but 
inefficient employees— managers as well as others— can be introduced 
as a specific limitation in the model. This permits the evaluation of 
the effect of such a policy in terms of profits. 

More than the employee's salary is involved; the opportunity to 
hire better employees is lost. One firm that made such a study decided 
to spend $25,000 to retire certain employees so as to improve the ad- 
ministrative process of approving sales commitments. Their study in- 
dicated that, by effecting the personnel changes, the company would 
save approximately $100,000 a year from reduced investments in inven- 
tory as well as from smoother production volume. 

Communication Network. The programming concept requires the 
establishment of a communication network which will transmit infor- 
mation rapidly and efficiently to the various levels of management 
where decisions are made. 

A number of companies interested in planning and control recently 
have initiated studies of their communication process. They have 
found that in their organizational structures often there are conflicts 
which make communication inefficient. Since this is true even when 
the process is judged in terms of needs that are less exacting than those 
which will be found in programming, the communication problem will 
no doubt be difficult to solve. 

For example, in a particular company the budget objective given 
by top management to its staff usually is sent on to operating levels for 
determination of their specific portion of the plan. But changes in 
objectives occur as time passes. As a result, those departments which 
depend on others for their data and thus make up their budget later, 



Programming, Scheduling, and Feedback 151 

tend to incorporate some of the newer objectives. This makes the 
budget inconsistent. Time pressures prevent ironing out all of these 
difficulties, so that eventually rough-cut methods are used to complete 
the budget in time for directors' meetings. 

Even if a better communication link could be established, two other 
obstacles must be overcome to achieve effective combination of the 
various planning factors in determination of the best program. These 
are the amount of arithmetical computation that must be performed 
and the difficulty of selecting the best alternative. 

Computation. The arithmetical computation often can be handled 
by an electronic computer. But before this can be done the relation- 
ships between the planning factors must be analyzed and reduced to 
an explicit, systematic form. This is not a simple process. Programmers 
have learned that one of the major dangers in their work is to assume 
that some relationships persist while others change. To take a time- 
worn example, the overhead rate used in cost computation is often 
related to direct labor cost. If technological change shifts the usage 
of labor, the old rate must be reexamined, which may involve a great 
deal of computation. More difficult examples involve the relationship 
between whole departments, or between product lines. In such cases 
a whole series of relationships exist, requiring a complete reexamina- 
tion when conditions change. 

Selection of Alternatives. Frequently it is difficult to select the best 
alternative course of action. There are always many possible pro- 
grams for accomplishing a given objective. The variations include not 
only different combinations of activities in the same fiscal period, but 
also possible postponements of various potential activities, such as the 
introduction of a new product. As a result, though the objective of 
planning and control may be to obtain the optimum profit consistent 
with sound, continuous growth and competitive conditions, in practice 
it has been impossible to determine when a company has reached the 
optimum profit. 

Optimum Allocation. Optimum allocation of effort and resources, 
as defined by business economists, calls for the use of marginal pro- 
ductivity analysis— increments of resource applied should produce in- 
crements of return. In other words, the last dollar spent in a given 
way should return at least as much in the given application as it might 
have done in any other possible application. But businessmen have 
found that the economist's analytical tools often are too limited for 



152 Electronic Computers and Management Control 

planning purposes. Businesses are not always able to make marginal 
adjustments. Plants or labor forces seldom can be increased or re- 
duced at incremental rates. It may not be possible to drop below- 
marginal product lines (marginal as measured on an individual prod- 
uct basis ) without over-all sales losses or without violating some oper- 
ating policy. 

Mathematical programming offers hope for improving on the arbi- 
trary standards of the "rate-of -return" method and for overcoming the 
difficulties inherent in experimentation with marginal adjustments. A 
program would forecast the optimum level of profits and make the 
necessary specific allocations of resources. Using such a program, the 
businessman could move directly toward his objective, making alloca- 
tions in the most efficient manner rather than attempting to progress 
by a process of trial and error. 

The very nature of mathematical programming as a concept may 
enable managers to reorient their thinking in more objective, measur- 
able terms. This already has proved to have considerable value in 
itself. It has been particularly useful to business groups that are in 
charge of plans for installation of electronic systems, and also for the 
designers who work for computer manufacturers. 

However, caution must be exercised. Much work remains to be done 
on all phases of program development. Many factors in business prob- 
lems as yet cannot be measured in quantitative terms. Many relation- 
ships appear to be nonlinear, and techniques for handling them still 
require further development. The human brain still must and presum- 
ably always will continue to be used to evaluate factors where informa- 
tion concerning them is incomplete. 

Development of the Programming Concept — Strategy and Tactics. 
Although experience is limited, and no full-scale programming projects 
have yet been attempted, it seems probable that the programming of an 
actual company will be undertaken in two major steps. 

The first step involves determination of long-run policies— the strate- 
gical decisions as to the over-all objectives of the company. These 
include: 

1. Determination of the profit goal (presumably profit maximiza- 
tion). 

2. Formulating and measuring the interrelationships between this 



Programming, Scheduling, and Feedback 153 

goal and other objectives, such as business statesmanship, product 
leadership, and so on. 
3. Integration of these objectives with operating decisions. 

The tactical decisions concern problems such as: 

1. Where the company will get its resources. 

2. Which new resources it should introduce into a going situation. 

3. Which resources it should shift to new products, and when. 

4. How best to use the resources available. 

5. What markets the company should enter. 

6. What share of the market the company should get. 

7. When the company should expand productive facilities. 

8. What personnel, public-relations, and social policies it should 
adopt. 

9. Whether the company should liquidate certain operations or 
expand in a new direction, and so on. 

Due to difficulties encountered in measuring the relevant factors, 
and in handling the relationships, so far the scientific techniques seldom 
can compete with strategy decisions made by business managers with 
experience and resourcefulness. But in spite of such limitations, man- 
agement scientists believe that it is at the policy level that the ultimate 
pay-off of new analytical methods will be greatest. Research in this 
direction is being urged upon top management. The concept is hard 
to sell. Results will not be immediate or certain. 

The second step, the tactical decisions, involves the establishment of 
shorter-run goals and operational plans which make optimum current 
use of the resources available. 

For example, the tactical program establishes the product mix for 
the period, the size of the labor force, the current personnel policies, the 
inventory policies, and the amount and type of productive resources to 
be used to make the various products. 

This policy concept also can be described in terms of "inputs" and 
"outputs." The desired "outputs" for each function of the company can 
be stated, e.g., dollar sales by products, the number of units to be pro- 
duced, etc. The inputs are the various kinds of resources to be ap- 
plied: advertising, sales force, personnel training, materials, machines, 
finances, etc. The inputs and outputs are so related that they fit the 



154 Electronic Computers and Management Control 

policies of the program and thus achieve the objectives of the company 
in an optimum way. 

Role of the Computer and Management Sciences. Communication 
networks of the type needed to furnish the timely information needed 
for programming which products to make, when to make them, where 
to sell them, how to make them, etc., have not been developed as yet. 
Electronic data-processing systems, along with the mathematical tech- 
niques, hold much hope for the future in this area. For example, it 
should be possible to handle budgetary computations more rapidly 
and much more effectively. 

An example of a tactical program is a decision model developed by 
the RAND Corporation for a refinery. The model gave the best prod- 
uct mix, considering the productive capacity, the forecast of market 
demand, and the technical operating limitations. As various forecast 
factors changed, the model could be run through a computer and a 
new optimal program selected rapidly. 

The time required to program part of an operation usually is not 
as important as the time required to effect a change in the over-all 
situation. For example, if winter comes early in the Northeastern 
states and the storage capacity there of a petroleum company is taken 
up by lighter products, delay might occur before fuel oil could be 
moved into the area, regardless of how fast the distant refinery could 
change its product mix. It is hoped that such considerations can be 
built into the model in the future to help make better decisions. 

Incidentally, mathematicians recently have suggested that such a 
problem might best be solved through "simulation." Simulation means 
approximation of conditions rather than precise quantification. Simu- 
lation has been used by engineers for many years, but as yet there has 
been little use of simulation for programming business problems. Elec- 
tronic computers can be used for simulation. The method offers con- 
siderable opportunity for worthwhile research. 

Now that digital electronic computers are available, for the first time 
it is economically feasible to handle the large mass of variables re- 
quired for budgetary programming. The computers can make logical 
decisions as well as do mathematical manipulations. Their logical de- 
cisions are somewhat comparable to many of those made by middle 
management— decisions which can be formalized. For example, a 
computer can be instructed that if a business budgetary program does 
not result in the realization of a given profit rate, the computer should 



Programming, Scheduling, and Feedback 155 

reject it and automatically consider a new program. The logical de- 
cision rules can be set up in the machine, if and when the specific 
programming factors of the business are quantified by management. 

The concept of programming seems to add new life to planning, 
however successful planning has been in the past. Programming makes 
it possible to interrelate many objectives. Company executives will 
be able to consider the effect of many programs rather than only one. 
The programs can take into account top-level policies as well as plans 
for best use of the resources currently available. 

The use of programming does not eliminate the need for competent 
managerial judgment at lower levels. In many cases a program can 
only set an objective and outline the way to reach it. The local man- 
ager or foreman must still work out the details, such as the approach 
to take to a particular customer or the rate to run a machine if the 
raw material is off-standard. 

Programming makes it possible to reexamine and evaluate concepts 
such as decentralization. It will provide for the delegation of certain 
decisions as well as establish a better means of setting standards, 
variations from which will let top management know how the de- 
centralized units are doing. At the same time, programming will 
emphasize the need for centralized consideration of the over-all objec- 
tives. 

However, there are many limitations which will have to be overcome 
before this concept of programming can become effective. Mathe- 
matical programming is not a cure-all which can replace management 
judgment at all levels. Currently available tools are only useful for 
particular types of programming, usually where costs or revenues can 
be seen as proportional to volume. 

Even if new tools are developed, the type of data currently gathered 
by business will have to be changed. In the past, accounting data were 
the major source of facts. Newer methods of programming will also 
rely upon engineering data, cost estimates, production quality, etc. 

Too frequently it happens that important factors, such as the com- 
pany organization and lines of communication, cannot be taken into 
account in present mathematical programming. The introduction of 
such factors into models requires design of new measuring units. 

Despite their speed, electronic computers may be limited in their 
capacity to deal with the requirements of some programming prob- 
lems. No matter how desirable the quantitative data, there may be no 



156 Electronic Computers and Management Control 

economical means of handling the sheer quantity of arithmetical com- 
putations. 

SCHEDULING 

Once the best program has been selected, it is necessary to establish 
a precise, detailed, yet flexible means for carrying out the program. 
This can be called "scheduling." 

Scheduling a program involves the physical details— quantities, quali- 
ties, locations, etc.— and the precise times when the various activities 
are to be completed, and by whom. In the past this has been accom- 
plished by middle-management decisions, aided by the established 
policies, the systems-and-procedures manuals, and by the budgeting 
process. 

Scheduling Requirements. Scheduling a program may involve the 
interrelationship of many factors. For example, if production is to be 
increased, the personnel department may have to add more people, 
the training department more instructors; capital assets may have to be 
increased, the accounting office may need a larger billing section, in- 
ventory policies may have to be changed, and so on. The new level of 
production may make it economical to increase subcontracting of par- 
ticular kinds of work, so that available machines can be put to work on 
optimal operations. 

Planning Factors. To schedule these changes requires the ability 
to estimate a large number of variables, to organize relationships to 
get the best use out of the available resources, as established by the 
program. 

It can be taken for granted that programs will not always proceed as 
planned. Scheduling is prepared to answer questions as to what should 
be done when unexpected machine breakdown occurs, or when mate- 
rials are not delivered, and so on. Through use of mathematical models 
for scheduling, the effects of rerouting production, delayed delivery 
schedules, and similar changes can be traced through to their ultimate 
effects on financial needs and profit expectation. 

Organizational Goals. As indicated in the previous discussion of 
decentralization, large companies today are usually made up of sev- 
eral or many semi-independent organizational units. Each unit tends 
to have its own operational and professional objectives which differ 
in part from the company's over-all objectives. Scheduling, therefore, 
must keep the organization oriented in the right direction as well as 



Programming, Scheduling, and Feedback 157 

attempt to achieve the optimum local use of resources. This is equiva- 
lent to saying that scheduling is concerned with the basic organiza- 
tional structure and with the fundamental factors of motivation at all 
levels, and with the operations in all divisions of a company. 

Programming will be carried on under the assumption that a certain 
scheduled level of performance can be expected from each functional 
group. This implies that each group will know what is required of it, 
when and how the job is supposed to be done. 

Information Flow. In addition to these original instructions, the 
functional groups need a flow of current information. The managers 
must be assured that work in preceding functions is being completed 
in time so that it will be ready for their schedule of operations, or they 
must be confident that new instructions will be forthcoming if it will 
not be ready. 

Thus scheduling implies more than an original set of instructions. 
It calls for a means of rapid communication between the operating 
groups as well as between these groups and the scheduling managers. 
If the planning factors can be interrelated in "real time"— that is, as 
events happen rather than after they happen— then managers can ad- 
minister their functions more confidently and more efficiently. 

Development of the Scheduling Concept. Development of the 
scheduling concept will take place both in the decision-model process, 
which concerns optimum allocation of specific resources to specific ac- 
tivities, and in the operational stage, which concerns administration of 
the activities so that they will be performed at the specified times, 
using the allocated resources. 

While some scheduling problems have been solved by the mathe- 
matical techniques, there still is no well-developed system which com- 
bines the various factors and produces the best schedule for the com- 
pany as a whole, one that integrates with the program. Not only is 
there a growing awareness of the need for such a system, but a number 
of researchers are combining various tools in the hope that an inte- 
grated mathematical scheduling process will result. 

Descriptive mathematical models have immediate applicability to 
the problem of stating the specific quantities and times at which specific 
resources are to be made available to specific activities. The sched- 
uling of production and the control of inventories are suitable appli- 
cations for such techniques. Mathematical simulation of processes 
often can be used to provide management with facts in a form which 



158 Electronic Computers and Management Control 

facilitates decisions. Resulting savings often are substantial, and in 
many cases much larger than clerical cost savings can be. 

To the extent that mathematical models can be developed to simulate 
the scheduling requirements, it will become more and more feasible 
to use electronics to make routine decisions. For example, employing 
a properly constructed model, a computer could be instructed to report 
inventory items which are below a minimum level, to examine the 
status of the need for those items in view of the current production 
schedule, and on that basis to determine whether a reorder should be 
placed. Given a currently maintained list of price quotes from sup- 
pliers, freight rates, time to deliver from suppliers' warehouses, etc., 
the computer could choose the best offer and print out a purchase order, 
ready for mailing. 

An area which has been explored to some extent is that of establish- 
ing purchase requirements. Models have been developed which show 
the purchase needs quickly and economically, using computers to "ex- 
plode" the bills of materials and to store the basic bills in an efficient 
manner. Learning curves have been used to set prices. 

Scheduling requires evaluation of the various ways of meeting the 
programmed activities. It usually is possible to produce a given line of 
products in different ways, using different methods and different rates 
of production. The best way of meeting the program is, therefore, part 
of the scheduling process. 

It appears that the growth of the scheduling concept will tend to 
change some managerial duties. Middle managers will be used less 
as processors of data, as people who make decisions largely according 
to rules and policies established by their superiors. They will be used 
more and more on borderline problems and on new situations which 
still call for human intelligence because they have not yet been suc- 
cessfully integrated into the model. 

Also, middle managers will have a chance to reestablish personal 
contacts with customers and employees. These are contacts which 
the managers have been losing as they found themselves more and 
more isolated in their offices while trying to cope with the streams of 
detail that flow to them in present systems. 

Mathematical models and electronic computers form a natural team 
for scheduling. Many scheduling models already developed are in a 
form well suited for solution by present computers. However, large- 



Programming, Scheduling, and Feedback 159 

size matrix models cannot be handled economically by present elec- 
tronic systems. It is hoped that future developments in logical design 
of electronic systems will lead to a solution of this problem. This means 
that special-purpose electronic systems may be developed which are 
economically preferable for scheduling, a possibility which has serious 
implications for managements that are choosing an electronic installa- 
tion. 

As the mathematical requirements for scheduling are obtained, they 
will assist in establishing the specifications for electronic systems. In 
time, electronic systems should be used to help schedule the remaining 
clerical routines, to relate the schedule for one business function to 
that of others, and also to provide a tool to facilitate evaluation of both 
the program and the schedule. 

FEEDBACK 

Recording, processing, and transmitting information concerning op- 
erations in a manner and in time so managers can control the operations 
is called "feedback." Feedback is a method of determining which data 
are relevant, how they should be obtained, and how they should be 
presented in order to accomplish the purposes of programming and 
scheduling. As with most system concepts, it is misleading to describe 
parts of the situation without relating them to the whole. Figure 5 
shows a simplified diagram of the role of feedback in the system con- 
cept under discussion. 

















PROGRAMMING 












i 




i 


■ 




1 




SCHEDULING 




FEEDBACK 








i 


r 










OPERATION 





























Figure 5 



160 Electronic Computers and Management Control 

Many present systems attempt to provide information, such as cost 
variances, directly to the managers in charge of operations. But sel- 
dom can they do so in a form or in time so that action can be taken dur- 
ing the course of the operation. As a result, managers set up their own 
little reporting systems, in an attempt to remedy the deficiency. 

Cost reports may be considered in schedule making and in planning, 
but they seldom are used in an integrated manner. It is too difficult, 
time-consuming, and expensive to trace through the system the many 
changes that result from rescheduling or replanning. Instead, resort is 
usually made to the "other things equal" assumption unless this is so 
obviously untrue that some recognition must be taken of the side effects 
of the changes. 

For example, if costs are too high on one product, a decision to 
change the method of production, or to raise the selling price, or to cut 
back production, etc., rarely gives full consideration to all the effects 
that the action will have on current operations, on other products, and 
in areas other than production. 

Feedback Requirements. The feedback concept emphasizes the im- 
portance of providing information which can be used to monitor both 
the program and the schedule. The monitoring can be automatic, as 
in the case of "automation" employing a servomechanism. It can be 
semiautomatic, employing dials, other instruments, or reports to initiate 
managerial or clerical reaction according to a fixed set of rules. Or it 
can be discretionary, as with an income statement given to a board of 
directors. 

Relevancy of Data. Feedback involves rapid evaluation of data 
from various sources. It requires determination of which data should 
automatically actuate controls and which should be passed on to the 
human link in the planning and control system. The character of the 
determination will change with time; as more and more automation 
is introduced the decisions left to humans will decrease in number 
but increase in relative importance. 

The problems involved in establishing a feedback system are con- 
siderable. As technology advances and as large organizations seek to 
decentralize, the problems increase. So far, the work done on a scien- 
tific basis by business in this field is inadequate. Pioneering has been 
led by groups such as those at the RAND Corporation, which have 
made laboratory studies of the way that information is sought and 
transmitted in complex organizations. The studies were made by teams 



Programming, Scheduling, and Feedback 161 

composed of mathematicians, statisticians, and psychologists, aided by 
neurologists and electronic engineers. 

Illustration of Relevancy. The only area in business that has been 
explored in this way, even to a fair extent, has been the cost system. 
It has been recognized that information produced by a cost system 
should be useful and produced in an economical manner. But too 
often the cost system is oriented to providing figures that the account- 
ant can use in costing sales and inventory, for published reports and 
tax purposes, with managerial uses provided only as a secondary fea- 
ture. A more complete description of the factors that should be eval- 
uated in developing a good cost system would include, as a minimum, 
such items as the need to provide information used in operating re- 
ports and financial reports, to value inventories, to fulfill accounting 
principle requirements, to meet tax or other governmental require- 
ments, to provide a basis for scheduling and other action taking at 
various levels of management, and to aid in programming the future 
activities of the company. 

Further, the data should be provided in a timely and economical 
manner. This means that clerical and other related processing costs 
should be at a minimum, that data should arrive in time so they can 
be used for control purposes, that the data should be consistent, that 
the cost system should avoid duplication of other data systems, and 
that the reports should provide maximum communication consistent 
with the amount of time for reading warranted by the need for the 
data, and so on. 

Data for cost accounting usually must be recorded as the products 
flow, otherwise the probability of incomplete recording is high. The 
criteria for selection of the proper cost centers seldom are established 
on a scientific basis, although, as indicated earlier, procedures studies 
have produced some advances in this area. But leading texts and hand- 
books still tend to define cost centers in terms of the need to apply 
direct costs and overhead to the product as it passes through certain 
groups of equipment. This computation not only is costly, but the 
overhead application in many cases is rarely of any consequence as 
far as managerial control of the operation is concerned. On the other 
hand, since the costs derived tend to be considered as "actual costs," 
businessmen often try to price sales in the light of the computed cost 
figure. 

Cost systems often become even more complicated and cumbersome 



162 Electronic Computers and Management Control 

when many of the costs are incurred at so-called "nonproductive" cost 
centers. The amount of computations may become tremendous. In 
the attempt to trace the costs through the products, several companies 
have used large-scale computers for cost distributions. 

It is evident that a cost system has several dimensions— financial- 
statement product costing, responsibility costing, project or "purpose" 
costing, etc. Joel Dean, in Managerial Economics, has given an ex- 
tended discussion of the different types of costs that may be of conse- 
quence for different types of decisions. Among these are: traceable or 
direct costs as opposed to common or indirect costs, variable costs as 
opposed to constant costs, book costs as opposed to out-of-pocket costs, 
incremental costs as opposed to sunk costs, escapable costs as opposed 
to unavoidable costs, controllable costs as opposed to noncontrollable 
costs, opportunity costs as opposed to outlay costs, past costs as op- 
posed to future costs, short-run as opposed to long-run costs, and his- 
torical costs as opposed to replacement costs. 

Since management has not been able to obtain cost data at will, in 
the manner best suited for a particular decision, many executives have 
tended to concentrate on one type of cost and encouraged their staff 
analysts to develop methods for determining that one particular type. 
Some management-consulting firms have built reputations through con- 
centrating on one or a few of these cost types. 

The one which seems to be favored at present is the cost- volume rela- 
tionship, shown through analysis of fixed and variable costs as "break- 
even" charts, or "P-V" relationships. These have proved helpful in 
many situations, and it is not expected that they will be abandoned. 
Rather, their use will be augmented by making available more of the 
several possible ways of analyzing costs. 

Generally speaking, data should go first to the lowest level of the 
organization, to the operating men. If higher levels get the reports 
first, they will try to take corrective action before the lower levels know 
the score, which inevitably creates friction. The "black book" system 
will be revived for protective purposes. 

Timeliness. Since programming is concerned with such matters 
as the optimum allocation of resources, as a rule the feedback for 
this purpose need be neither overprecise nor rapid. Programming 
plans are seldom developed overnight. A decision to enter a new field 
with a new product line may require long periods of research, lengthy 
negotiations with new distributors, or drawn-out negotiations for the 



Programming, Scheduling, and Feedback 163 

purchase of going concerns. And since the program is judged by such 
standards as rates of return, substantial rounding off of figures or some 
imprecision of estimates may be quite satisfactory. 

Scheduling requires faster feedback. If a suitable model has been 
developed, it may be possible to reschedule at short notice in order to 
meet the exigencies caused by breakdowns, supply failures, short-notice 
customer demands, and so on. The frequency of such emergencies 
will vary greatly from business to business; a job-order shop will re- 
schedule more frequently than a firm with a few standard products 
for which the demand is relatively steady. Nevertheless, even where 
the product is seldom changed, research groups have been studying 
the possibility of improved scheduling of production and warehouse 
distribution as a direct reflection of current sales activity. Cross- 
country communication networks have been established. 

The feedback to the operating managers obviously must be imme- 
diate, for optimum control. If possible, it should indicate the imme- 
diate future trend. Managers would like to have information as cur- 
rent as the minute-to-minute rate of sales in various departments of a 
store, with the implications for sales the rest of the day and week, if 
this could be obtained economically. A production manager would 
like to have a constant flow of information on the movement of partly 
finished materials from other departments to his, and on the flow of 
production past the key check points of his operation. 

In many cases managers attempt to achieve this goal now by rough- 
cut means. Sales managers "walk" the store, and through experience 
are often able to gauge sales activity very closely. Production man- 
agers watch bottlenecks, such as a press past which most of their 
production flows. They are able to estimate how the line is doing 
through observation of the work waiting at the bottleneck. 

A good feedback system would achieve the same results, as a regular 
part of the procedure. 

The speed demanded becomes greater as one goes from program- 
ming through scheduling to operating needs. At the same time the 
amount of information required tends first to rise, then to decrease. 
A few key figures, reported rapidly, often are enough to tell the operat- 
ing manager how his shop is getting along. 

Reporting. While feedback reports to operating managers deal 
primarily with immediately current activities and short-range fore- 
casts, those for scheduling require some consideration of forecasts for 



164 Electronic Computers and Management Control 

longer periods. Reports for programming are concerned almost en- 
tirely with such estimates, which may be for years ahead. 

Sometimes the programming needs for data are said to be like look- 
ing at the organization as though it were a system of flows to and from 
a series of "black boxes," without worrying too much about what goes 
on inside the black boxes. This concept may be helpful if it is remem- 
bered that the "flows" include personal relations, attitudes, and so on, 
as well as products and dollars. Scheduling worries more about the 
inside of the black boxes and the timing of the flows. 

Accuracy. Another difference deals with the relative accuracy of 
the data reported. In many cases it is enough for the operating man- 
ager to know that something is going off schedule and an indication 
of how much he should do to correct the aberration. If the variance 
is fairly sizable, it will probably have to go out of his hands anyhow, 
in the sense that rescheduling may be called for. On the other hand, 
programming may be carried on quite successfully with estimates that 
are only expected to lie within a range of several percentage points. 
Many sales managers, for example, would be glad to settle for the 
assurance that a sales forecast would be within five per cent of accu- 
racy almost all the time, especially if they also knew that the system 
of feedback would indicate how the forecast should be changed as 
current sales reports start to flow in. 

Scheduling requires more accurate measurements. Such items as 
the current status of the operations are needed in order to check out the 
schedule. For example, if the count of parts is inaccurate and the 
actual amount too low, then completed items cannot be assembled in 
the scheduled quantity. 

Development of the Feedback Concept. In recent years much pub- 
licity has been given to the concept of management by exception. 
While this has frequently proved to be an improvement over previous 
systems in cutting down the detail brought to the attention of man- 
agers, a new difficulty has arisen. In his attempt to get the operation 
back to standard, too often the manager overcompensates or under- 
compensates for the variation. A classic example is the attempt to keep 
costs in line by the cutting of advertising allowances when sales fall. 

The problem not only is to report that an operation is out of line, 
but to show the relative degree to which it is out of line, together with 
an indication of the amount of corrective action which should be taken. 
At present this corrective action usually is decided upon by the man- 



Programming, Scheduling, and Feedback 165 

ager largely on the basis of his accumulated personal experience. As 
more is learned about the factors which throw operations out of line, 
these corrections can be incorporated into a feedback system so that 
they will become more and more automatic. 

This increase in understanding has already started at the operating 
level, with automation. So far, little knowledge exists with which to 
set standards at the top-management level. As indicated earlier in 
this chapter, a rate of return on investment has been used to some 
extent for measuring top-management performance, but the shortcom- 
ings of the method are such that the need exists for more scientific 
standards to be used for this appraisal. 

Feedback puts prime emphasis on the usability of information. This 
means that the information not only must be what the managers can 
understand, but also it must be directly relevant to their individual 
needs and arrive in time for them to use it for current decisions. For 
this reason, as indicated at the start of this section, feedback can be 
studied only in the light of programming and scheduling concepts. 

If accounting is to be relied on to provide the data for scheduling, 
some changes in present systems usually will be required. For ac- 
counting to provide the necessary accuracy, the process usually fol- 
lowed at present calls for so many checks and rehandling of the figures 
that the data are obsolete when they arrive. If the data can be meas- 
ured by counters, either mechanical or electronic, and transmitted 
along electrical circuits, or through the use of such devices as indus- 
trial TV, then the scheduling process will be aided considerably. 

Reports of data in tabular or chart form may be suitable for program- 
ming, but inadequate for scheduling. Middle management has to 
relate data to much other specific information in order to make a de- 
cision. If the data can be brought together and processed in terms of 
a mathematical model, the decisions obviously would be greatly facili- 
tated. The feedback would automatically provide a constant testing 
of a suggested course of action. 

Feedback is intimately concerned with both the organization struc- 
ture and the method of reporting to each level. Although management 
today frequently feels that it receives too many reports, in almost every 
instance a justification can be found for supplying the information 
because of some type of decision which occasionally has to be made. 
Only if the manager could be sure that the particular bit of informa- 
tion found on an infrequently used report would be readily forthcom- 



166 Electronic Computers and Management Control 

ing on requests would he cheerfully agree to elimination of the regu- 
lar report. 

As a rule it probably will be found that the nearer the operating 
level, the more the reports are concerned with physical quantities, 
while top-level programs will more often be in dollar terms. Schedules 
tend to be primarily physical, as they pertain to departments like pro- 
duction, but may take on dollar aspects in all areas. Some schedules 
may be wholly in dollars, however, as in forecasting cash requirements. 

It should be apparent that electronic systems, and use of mathe- 
matical models, will be of considerable assistance in providing better 
feedbacks. Speed, accuracy, mathematical manipulation, condensa- 
tion, reporting of exceptions, and so on, are precisely the sorts of abili- 
ties they offer. Since computers can be instructed to handle data in a 
variety of ways, it is possible to arrange for different types of reports 
to be made to different levels of management, starting from the same 
set of original data. Such systems will be discussed in Chapter 9. 

SUMMARY 

Programming, Scheduling, and Feedback. Electronics has already 
been recognized as a means of improving the communications which 
are necessary for present systems of planning and control. Don G. 
Mitchell, chairman of the board of directors at Sylvania Electric Prod- 
ucts, has stated in a speech describing new electronic developments in 
the field of communication: 

The many benefits of this [decentralization] policy include far-greater 
flexibility in operations, faster action, new and better ideas, and greater 
efficiencies all along the line. This obviously creates quite a problem 
in communication— so that the decentralized plants will not be operating 
in separate vacuums, like electrons that have escaped the force binding 
them to the nucleus of an atom. 

Although the separate plants are decentralized in authority and re- 
sponsibility, and are given a job to do and permitted to do it, they are 
nevertheless part of a whole. You can carry "state's rights" too far in 
industry. To assure coordination— not interference with a local manager's 
prerogatives— companywide information must be gathered speedily and 
delivered to those in managerial and policy-making positions in time to 
do something about it. The accountants call this integrated data proc- 



Programming, Scheduling, and Feedback 167 

essing— which is a fancy way to describe the process of recording and 
processing day-to-day information, and presenting it rapidly in summarized 
form so that management can act while the situation is fresh. 

Here again is a major opportunity for the accountant: helping to de- 
velop this system of high-speed communication, deciding what information 
is required, and what should be done to that information to obtain co- 
ordinated reports. Those who have had experience with decentralized 
companies may wonder if such a system would not act to the detriment 
of the operating man, and tend to centralize controls. It not only would 
not destroy decentralization, but would foster it by supplying the operat- 
ing organization with the required information, and the top management 
with the data it needs. 

The ways in which electronics can aid in this effort to achieve 
faster communication have already been indicated to some extent in 
earlier chapters. 

In addition, some of the newer mathematical tools can help develop 
methods to judge more rapidly the effectiveness of various alternative 
management plans. These methods can take into consideration many 
different ways of doing things, ways which explicitly introduce alterna- 
tive courses of actions. The analysts can evaluate the objectives of the 
company— profit and otherwise— and help select that alternative which 
will result in the best course of action. 

As Henderson and Schlaifer declared in the Harvard Business Re- 
view: 

In recent years mathematicians have worked out a number of new 
procedures which make it possible for management to solve a wide variety 
of important company problems much faster, more easily, and more ac- 
curately than ever before. These procedures have sometimes been called 
"linear programming." Actually "linear" describes only one group of 
them; "mathematical programming" is a more suitable title. 

Mathematical programming is not just an improved way of getting certain 
jobs done. It is in every sense a new way. It is new in the sense that 
double-entry bookkeeping was new in the Middle Ages, or that mechani- 
zation in the office was new earlier in this century, or that automation in 
the plant is new today. Because mathematical programming is so new, 
the gap between the scientist and the businessman— between the researcher 
and the user— has not yet been bridged. Mathematical programming has 
made the news, but few businessmen really understand how it can be of 
use in their own companies. 



168 Electronic Computers and Management Control 

The purpose of management planning and control is not merely to 
speed communication, to reduce clerical costs, or to establish the de- 
cision rules now made by middle management. The purpose is to 
achieve the full range of the company's objectives. 

Many executives now believe that the key to full achievement will 
be the adoption of a new conceptual framework. It is their belief that 
only by a change in thinking can the gap be bridged between the 
electronics engineer, the mathematician, and the businessman. 



CHAPTER NINE 

Integrated Business Systems 



Whether a company has started with a single application for an 
electronic system or with a study of management planning and con- 
trol, the result usually has been recognition of the need for extending 
the work. For example, applications of electronic computers to pay- 
roll data processing have been expanded to include labor reports, cost 
accounting, etc. Studies of the scheduling aspect of the management 
planning and control system have led to the need for feedback. 

The term "integrated business system" combines the concepts of 
business data handling and of programming, scheduling, and feed- 
back. In this sense it covers both the electronic data-processing sys- 
tem and the system of management planning and control. Integration 
means that each is developed in full recognition of the capabilities and 
requirements of the other. 

An integrated business system links the event that originates an item 
of information with the events that occur wherever and whenever 
someone uses this information. The data flows can be traced directly 
from the point of original record until they reach processing or de- 
cision points. The flows then take one of two forms— combining or 
branching. These forms can be represented as follows: 

1. Combining: 

Event — > Original record — > Information flown 

— » Processing — » Report 
Storage of information — > Information flow ' 

Typically, a report combines items of information, of which one 
may be from storage. The storage item may be as simple as the 
product or employee code number identifying the object or per- 
son to whom the event occurred. Combining also may bring to- 

169 



170 Electronic Computers and Management Control 

gether two original records, or two items from storage. (A report 
is a form of storage, in the sense considered here. ) 
2. Branching: 

[— » Choice 1 — > Event 
-> Choice 2 -> Event 

Branching signifies the use of data by a manager to make a 
choice between alternatives. His decision programs or schedules 
a new event. 

In the typical business situation several items of information usually 
are combined on one report. Or, a choice is made from more than two 
alternatives. By analyzing business operations into these two basic 
forms they can be seen in their fundamental aspects. When these 
have been established, the elements can be synthesized into an inte- 
grated business system. 

THE INTEGRATED BUSINESS SYSTEM CONCEPT 

An integrated business system can be viewed as a communication 
process. Transactions occur, are recorded, and the record communi- 
cated or stored. Decisions are made by managers who, knowing the 
established policies of the company, consider the facts presented to 
them. When an individual manager makes a decision, he transmits 
information to others in the organization. They act, or use the infor- 
mation in turn as the basis for their own decision making. 

For example, when a machine breaks down, the worker usually in- 
forms the foreman. The foreman must make a decision from among a 
number of alternatives. He may relocate the work. He may request 
immediate repairs from the maintenance department. If the break- 
down is sufficiently serious, the foreman will notify the superintendent 
and the scheduling office. Someone will then reschedule. Meanwhile 
maintenance must decide whom to send on the repair job, giving con- 
sideration to other calls for their services. Later, the engineering de- 
partment may collect data as to frequency of machine breakdown. 
A report may be prepared for consideration by managers, who may 
decide to replace the machine. 

In an integrated business system, three activities can be distin- 
guished: 



Integrated Business Systems 171 

1. Information flow, direct transfer. 

2. Information process, report combining. 

3. Information analysis, decision branching. 

The first part of this chapter will discuss the problems involved in 
direct transfer of information. Information analysis, decision branch- 
ing, was covered in Chapters 6, 7, and 8 and will only be touched on 
here. The latter part of this chapter deals with the combining proc- 
ess and report creation, the integrated data-processing system. 

DIRECT TRANSFER OF INFORMATION 

A business is constantly transferring information from one place to 
another, starting with a number of points of origin and following 
through to a variety of end-users. Viewing this process objectively, 
it can be broken down into these basic elements: 

1. Originator of the information. 

2. The physical means of communication and the channel the mes- 
sage follows. 

3. Symbols used to convey the information (language, numbers, 
etc.) 

4. The message. 

5. User of the information, or storage point. 

6. Continuation of the process, in the sense that the user originates 
a new message, or the information later is drawn from storage. 

Communication often follows "loops," since some of it returns to the 
originator. Perhaps the return is only a notice that his message has 
been received. Often there is additional information in the form of 
instruction, or there may be variance comparisons with stored infor- 
mation. 

There always should be additional information flowing at some time 
from each user. Unless the recipient eventually takes some sort of 
reportable action, the original information must be considered useless. 
The communication process has failed. 

Because of their interrelationship, these basic elements will be dis- 
cussed in groups, rather than as separate items. 

Communication, Symbols. There are many kinds of information 
channels. People talk directly, use the telephone, or use other media. 



172 Electronic Computers and Management Control 

They send written reports, which often go in multiple copies to several 
parties. The information may be transferred in whole or in part 
through several levels of management. 

The paths followed may be formal or informal. By "formal" com- 
munication is meant the channels established in manuals, which de- 
scribe the type of information that shall be reported, who is to give it 
to whom, when it is to be reported, and so on. Standard forms usually 
are prescribed. The formal communication plan is closely related to 
the organization chart. 

Although a company frequently makes an effort to increase the 
relative percentage of information that is formally recorded rather 
than relying on informal communications, nevertheless the formal sys- 
tem has limitations. Not all contingencies or needs can be determined 
beforehand. Written reports may not impart a "feel" for a problem. 

It is hard to estimate how much of the information flow in a business 
moves through personal contacts in conferences, at lunch, in car-ride 
pools, at the water cooler, over the telephone, and so on. Some man- 
agers believe that it is well over half, and a few make it even higher. 
Some boast that they never read a written report. 

The electronic system can handle both formal and informal com- 
munication. It may be used to increase the percentage of the total 
information that is formally communicated. By making the formal 
communications network speedy and obviously superior to the in- 
formal one, it will tend to monitor and replace the less efficient medium. 

There are numerous drawbacks to the informal system, not the least 
of which is that it is inaccurate, especially if more than one transfer 
is involved. Anyone who has traced a bit of information as it passes 
verbally and informally from person to person knows how it can be- 
come distorted— figures change, and sometimes the meaning is entirely 
reversed. The successes of the grapevine tend to be remembered, and 
its many failures forgotten. 

However, the informal system is hard to displace and impossible to 
eliminate. Another possible improvement lies in making the informal 
system more direct, e.g., through industrial TV. 

This means that the barriers to formal communication must be ob- 
jectively examined and provision to overcome them incorporated as 
far as possible into the system itself. Among these barriers, the follow- 
ing can be distinguished as fundamental: (1) semantic differences, 
(2) different frames of reference, (3) status distance, (4) geographical 



Integrated Business Systems 173 

distance, ( 5 ) need for self-protection, and ( 6 ) pressure of other work. 

1. Semantic Differences. Although the problem is as old as time, 
it is only in recent years that businessmen have discovered the useful- 
ness of the scientific attention which has been directed to the semantic 
differences that exist within a language. An industrial combination is 
made up of a number of functional groupings. Each function— sales, 
production, finance, purchasing, product engineering, research and de- 
velopment, and so on— has its own set of terms, its own slang, and its 
own attitudes. Moreover, the problem of communication can exist 
within a function, when a small group develops an interest in a new 
field foreign to the experience of the majority. "Linear programming" 
has occasioned this sort of difficulty in a number of firms. 

Perhaps even more vital is the fact that all humans must use language 
and mathematical symbols with which to perform the act of thinking 
itself before anything can be communicated. Since all of us absorb 
our knowledge of language in many ways, the result has been that our 
very thinking processes reflect many unscientific beliefs and prejudices. 
These prejudices concern not only such general topics as religion and 
race, but more fundamental concepts such as the structure of language 
itself— the meaning to be attributed to the whole class of nouns, the 
significance of the grammatical construction of sentences, logical analy- 
sis, and so on. A number of books have appeared which indicate the 
nature of these difficulties, such as Ogden and Richards' The Meaning 
of Meaning, Percy Bridgman's The Logic of Modern Physics, Korzyb- 
ski's Science and Sanity (the latter popularized in several well-known 
works by Stuart Chase), and several publications by Rudolf Carnap. 

A thought expressed in symbols can only bear a relation to reality; 
it can never be the reality itself. A symbolic representation of this 
relationship is all that can be communicated. The communicated idea 
can hardly be less fuzzy than the thought that originated in the first 
observer, and almost inevitably something is lost each time the message 
is transferred. 

The importance of this problem has been hidden by concentration on 
other difficulties in the communication process, such as the overcoming 
of distance. But the communication problem will not be appreciably 
lessened without consideration of this basic difficulty. Even if two 
businessmen live together, have most of their major interests in com- 
mon, work together on projects, like each other, and so on, they still 
may not be able to communicate effectively. 



174 Electronic Computers and Management Control 

While we are far from the solution of this difficulty, it is a major 
accomplishment even to recognize the nature of the problem. Con- 
siderable resources are now being devoted by universities to the study 
of semantics, and business has at least become aware of the need for 
further scientific investigation in this area. 

2. Frame of Reference. Unfortunately the basic language barrier is 
only the beginning of the problem. Although the original statement 
may be clearly and correctly expressed, in so far as both the originator 
and recipient are concerned, each recipient has a different environment, 
a different "frame of reference," and even though he "understands" the 
communication he will never react to the information exactly as the 
first party did. Too often, this difference will result in substantial 
distortion. 

For example, in one company a production department thought that 
the purchasing department was not providing necessary parts on time. 
The belief persisted in spite of the information to the contrary reported 
on a parts shortage list. The production department's reaction to the 
situation was that purchasing simply didn't understand their problem. 

The same set of facts may be interpreted quite differently by various 
groups. In many companies, when sales fall off, the production depart- 
ment may want to increase output in order to lower unit costs so prices 
can be lowered, the treasurer may want to curtail production so as to 
conserve cash, the sales department may want to increase the advertis- 
ing budget, while the general manager may consider dropping the 
line . . . and so on. Knowing this diversity of interpretation, depart- 
ments frequently conceal or delay transmission of information. This 
point deserves separate discussion ( see point 5 ) . 

S. Status Difference. Effectiveness of communication in an organi- 
zation frequently bears a direct relationship to the difference in status 
between the parties involved. Managers and workers on approxi- 
mately the same level often communicate freely. Where status differs, 
communication often is better downward than upward. Devices such 
as suggestion boxes are useful but makeshift solutions to this problem. 

The personality of the executive has considerable influence on this 
situation. Executives who "walk their shop," keep their desks out with 
those of workers on lower levels of authority, or maintain "open-door" 
policies in their office can affect the informal communication pattern 
considerably. However, the process is time-consuming and is not al- 
ways desirable. 



Integrated Business Systems 175 

Formal communication between levels is difficult to influence. De- 
spite the old saying, bad news does not travel fast in an upward direc- 
tion unless deliberate attention is given to making sure that it will do 
this. A major function of the internal audit staff in many companies is 
to minimize attempted distortion or deliberate falsification of reports. 

As an example, the "idle time" concept is highly thought of in text- 
books, but in practice it is difficult to monitor because of obvious con- 
notations. The effort to avoid such a report may lead to considerable 
distortion. In one large company the present practice of a certain 
group is to report all effort 4 days late, so that idle periods can be 
covered up. 

Incidentally, this practice has been uncovered in Russia, where a 
few division leaders have set their yearly quotas at the level of the 
previous year's production. They then reported a year late, holding 
over net balances in inventory. 

4. Geographical Distance. Decentralization of management and 
expansion of facilities in various parts of the country or the world have 
made the coverage of distance an important factor in communication. 
This part of the problem, in many respects, has been solved the best, 
perhaps because it appears so obviously as a problem. Some executives 
have learned "to live on the telephone" almost literally. 

Nevertheless communication of large volumes of information over 
long distances is still expensive and not always as effective as it should 
be. Companies still find it worthwhile to hold meetings of their sales- 
men, and of various levels of the executives, to supply the spark that 
is lacking in formal communication. 

5. Self-protection. As indicated earlier, if given the opportunity, 
people usually slant reports so as to minimize information which puts 
them in a bad light. Knowing this, many executives auomatically in- 
crease the significance attached to any piece of bad news and tend to 
discount good news. The persons who report become aware of this, 
and so they increase the degree of minimization of bad news, leading 
to further reactions by executives, until the matter gets out of hand. 

There is no way to eliminate this human tendency, but some attempt 
must be made to stabilize the influence so that it is held within tolerable 
limits. An effort to bypass the human reporter wherever possible, 
through use of automatic recording devices, is being made and will 
probably increase in the future. 

The points of information origin must be chosen with considerable 



176 Electronic Computers and Management Control 

care. Safeguards must be built into the system to make sure that the 
operating personnel do not slant their efforts toward the check points 
and avoid fulfilling other obligations that supposedly were to be accom- 
plished simultaneously. A simple example would be the efforts of 
production men to get items past inspection, regardless of the other 
consequences of their efforts. 

6. Work Pressure. Most executives are judged by the quantity of 
work they turn out and not by the quantity or quality of the informa- 
tion they originate. (An exception may be special reports which they 
occasionally make for superiors.) Communication may be slighted. 
As work pressure increases, communication is delegated more and more 
to clerical staffs, with a minimum of interpretive material added by the 
executives. 

In some firms the situation becomes so bad that it is formally recog- 
nized. Representatives of the data-processing departments, under the 
controller, are sent in. 

Recording may be in the hands of an outside group, but unless their 
efforts are trusted by the operators the resulting reports will cause noth- 
ing but friction. Some firms have attempted to solve this by having 
the recorders responsible to both the operating manager and the func- 
tional head of accounting. In any case, the matter remains a problem, 
correspondingly greater as work pressure mounts. 

Originator, Message, and User. The design of an integrated busi- 
ness system should consider simultaneously the position of the origi- 
nator of information, the form of the message, and the need of the 
user. This is a difficult task. 

The closest approach in present practice to the type of analysis which 
is required is found in studies of decentralization. An example is the 
investigation undertaken by H. A. Simon, G. Kozmetsky, H. Guetzkow, 
and G. Tyndall for the Controllership Foundation (published in 1954 
as Centralization vs. Decentralization) . 

This study defined an administrative organization as centralized to 
the extent that decisions are made at relatively high levels in the organ- 
ization, and decentralized to the extent that there are important delega- 
tions of discretion and decision making from higher to lower levels. 
After separating management's use of figures into three categories— 
(1) score card, (2) attention direction, and (3) problem solving— the 
study found that the three different uses are not of the same importance 
to executives in different parts of the operating organization. At all 



Integrated Business Systems 177 

levels of management, score-card and attention-directing uses of in- 
formation are more frequent than problem-solving uses. Problem- 
solving uses of accounting data are encountered most frequently in 
staff units that have responsibility for making special studies. 

Because of these differences in the persons to whom the information 
is directed, and in the content of the information, a controller's depart- 
ment that is well organized to provide one service may not be well or- 
ganized to provide another. The authors concluded that for score card 
and attention directing it is important for the controller's department 
to develop a close and direct relationship between accounting and 
operating personnel. 

Other authors have gone so far as to advocate completely autono- 
mous reporting groups, often comprised of only one or a few people, 
each reporting directly to and for an operating manager. 

For problem solving there are two possible solutions. One is for 
the controller's department to make more elaborate reports so that the 
information needed will usually be found on the regular reports. The 
other is to devote more attention to special analyses that go back to 
the raw data, to build up specific figures that are needed to answer 
specific questions. The authors of the controllership study recom- 
mended the latter course. 

Since this study was based upon investigation of actual operations 
in a number of companies, it is evident that some businessmen are al- 
ready moving in the direction of fitting the data-processing system 
more directly to the needs of the users. If score-card and attention- 
getting data are reported as gathered directly to managers, and if 
problem-solving data are developed from raw data by special staff 
groups, it is evident that for the first user there would tend to be report- 
ing of data only as required. The "time-schedule" problem of data 
processing would be minimized. 

However, if the data must be used for several purposes, as is usually 
the case ( for payroll, etc. ) , the duplicate processing required often be- 
comes expensive. There are a number of ways that this "message" 
problem can be relieved. 

The simplest is to report the data in physical terms where possible. 
Dollarizing the data is time-consuming, expensive, and frequently un- 
necessary. Some items, such as manpower status, may be handled 
best by clerks reporting directly to superintendents, as straightforward 
counts. 



178 Electronic Computers and Management Control 

Where more complicated reports are necessary, such as of variances, 
it still may be possible to provide meaningful data as physical counts. 
If accumulations through time are needed, then the use of counters 
and of electronic systems may be indicated. However, this raises 
serious problems concerning the location of the equipment. Some com- 
panies are following their decentralized pattern, and ordering several 
smaller sized computers for each of their branches; some are consider- 
ing or have already ordered one or a few of the large-scale computers 
for their centralized record-keeping. 

Use of the computer, with its ability to process data rapidly, should 
not be permitted to encourage managers to ask for data they do not 
need. In many cases, if a model can be constructed of a particular 
operation, the manager can operate in terms of key inputs, outputs, and 
variables in his model. 

Eventually, as a number of these models of parts of the system are 
developed, the time may come when development of the over-all model 
of the system can be attempted, with reasonable hope of success. 

Information Analysis. Little work has yet been done on the study 
of the integrated business system so as to incorporate decision mak- 
ing— "branchin g." 

The interrelationships of decision making are difficult to establish, 
especially in view of the large part that informal communication now 
plays in the process. There have been some studies in this field, such 
as the work at RAND, at the University of Michigan, and at the Har- 
vard Business School on the functioning of committees. The Ford 
Foundation has granted Carnegie Institute of Technology a substantial 
sum for basic research in this area. 

Another difficulty in modeling the decision process is to indicate the 
role played by factors external to the company. 

Some work has been done in production control, employing both 
mathematics and electronics. Tentative models of systems have been 
constructed, but they have not been tested, nor has there been much 
attempt to integrate them with the rest of the system. 

Development of fully integrated business systems so as to provide 
formally for decision-making and the data-handling system is depend- 
ent upon research of the type described in Chapters 6, 7, and 8. Mean- 
while, present electronic applications usually employ the "combining" 
process. 

Many problems must yet be solved in order to develop integrated 



Integrated Business Systems 179 

systems for the combining aspect of the system— especially for sum- 
marization and reporting of information. Solution of these data-han- 
dling problems is a prerequisite for eventual combination with branch- 
ing systems. 

Development of the data-handling system for programming, sched- 
uling, and feedback will require precise determination of managerial 
requirements for decision information. It must take into account the 
action that flows from information, the manner in which the formal 
organization is structured, the informal organization of the company, 
and so on. 

Integration of this data-handling system with programming, sched- 
uling, and feedback must wait for the future. Equipment— electronic 
or electromechanical— to implement a fully integrated business system 
is not yet available. But the direction that developments are taking 
can be foreseen, and changes can be made in advance which allow for 
partial adoption of the new methods and techniques. This preparation 
will be aided by the fact that electronics and mathematics can be used 
right now to cut clerical costs, to aid in making better decisions, to 
study ways of organizing effectively, and for other similar purposes. 

Many companies already have taken the first step toward fully inte- 
grated business systems by installing integrated data-handling systems. 
Therefore the remainder of this chapter will be devoted primarily to 
discussion of the combining process. 

PREPARATION FOR INTEGRATED DATA HANDLING 

The problems created by inadequate business data-handling systems 
are well known. They arise from such items as the following: 

1. Useless and outmoded communications and reports. 

2. Duplicate records, created and maintained in the same organiza- 
tion. 

3. Storage of elaborate records to support claims of accomplish- 
ment, or in anticipation of questions that may never be asked. 

4. Excessive analyses and statistics, employed as substitutes for in- 
dividual judgment. 

5. Poorly designed records which are basically necessary, but are 
encumbered with extraneous and costly information of little use 
to anyone. 



180 Electronic Computers and Management Control 

6. Information assembled to satisfy individual curiosity, having no 
practical value. 

Most companies continuously study their organization, their infor- 
mation needs and methods of handling, processing, and reporting data. 
These studies have resulted in measurable economies and a degree of 
control over methods employed in the clerical, accounting, and ad- 
ministrative services. 

The approach in these studies usually has been in terms of "the ques- 
tioning attitude." Analysts have worried about such things as the 
proper way to view the functions involved, the proper way to categorize 
and relate the operating supervisors, the documents and reports, and 
so on. Or, time studies have been made and standards established, 
much like those found in the factory. Flow charts have been prepared 
to help determine the best way to process documents. 

In most cases the recommendations of analysts have had to be "sold" 
individually to each level of the supervision and management con- 
cerned. As a result, the recommendations have often tended to become 
shaped so as to facilitate this persuasion. 

In an integrated data-handling study, on the other hand, the over-all 
objective is not merely to reduce paper- work costs but rather: 

1. To increase management effectiveness through analysis of the in- 
formation needed for decision making and through the means of 
processing it in a timely fashion. 

2. To establish a continuous system, one which promotes greater 
coordination throughout the company— an interdepartmental point 
of view. 

There has been a tendency for business-system analysts to become 
diverted by the cost-reduction opportunities. Often they lose sight of 
these major purposes. Sometimes they forget that the best solution 
may be to stop collecting the data, rather than to process it better and 
faster. 

The availability of an electronic computer does not solve these prob- 
lems. In fact, it can compound the difficulties. The speed, versatility, 
and logical capabilities of electronic equipment enable it to produce 
almost limitless reports. But an increase in the volume of the reports 
does not ensure better decisions. 



Integrated Business Systems 181 

Voluminous detailed reports can confuse or overwhelm the recipi- 
ents. Some executives will fail to act because they deliberately choose 
to ignore the mountain of facts. Others will become uncertain because 
of the excessive amount of time required to sort out the significant data. 

On the other hand, elimination of needed information, through over- 
emphasis on clerical cost reduction, can hamper executive effectiveness 
to an equal if not greater degree. 

Electronic computers are an aid to those who wish to integrate data- 
handling business systems. But by themselves they cannot determine 
when a report is needed, or if it is useless and outmoded, or that dupli- 
cation exists as between two organizations. Computers are tools which 
can be used to achieve integration, but the user must first establish 
conceptually the system of planning and control. Then he can design 
the data-handling system and fit the electronic devices to it. The re- 
verse approach, fitting the electronic tool into existing data routines, 
is almost certain to achieve something less than an optimum result. 

Adoption of some concept, such as that of programming, scheduling, 
and feedback, is basic. The establishment of some fundamental con- 
cept is necessary to structure the system study. Only in this way can 
consideration be given to all end purposes of the data-processing sys- 
tem—not only to the need to prepare financial reports for stockholders, 
bankers, tax returns, and so on, but also to the specific needs of various 
levels of operating managers. 

Integrated Data-handling Studies — Experience to Date. There are 
two major classes of integrated business data-handling systems: 

1. Integration of the form of source data. U.S. Steel's "Common 
Language" Project is an example. Du Pont, SKF, and others are 
also developing their own systems. These try to integrate the 
source data into "one common form"— a punched tape or mag- 
netic tape— so that it can be used by different machines and for 
various functions, e.g., cost accounting, production control, billing, 
etc., without retranscription. 

2. Geographical integration. Information generated by various geo- 
graphical units— plants, sales offices, warehouses— are integrated 
into the business-data system. Alcoa has sales offices transmit to 
Pittsburgh, by teletype on punched-paper tape, all the sales orders 

received. Pittsburgh in turn assigns and transmits the orders to 
various plants, all within one day. 



182 Electronic Computers and Management Control 

Most companies have confined their efforts to part of the first class 
of problems— to integration of data requirements for a particular func- 
tion. Some have also undertaken the second class— geographical inte- 
gration. The number is growing rapidly. 

Most current studies tend to emphasize determination of the best 
equipment, and only secondarily the need for systems procedure 
redesign. However, some companies, like du Pont, SKF, Alcoa, and 
U.S. Steel, have undertaken major revisions of their data systems. U.S. 
Steel's integrated data program stresses three major objectives: 

1. Data should be recorded mechanically at the point of origination. 

2. Data should be processed mechanically without need for manual 
recopying of the information. 

3. The data system should be integrated to prepare for subsequent 
application of electronic equipment. 

In order to accomplish these objectives, U.S. Steel and other com- 
panies have selected equipment on which they record the original data 
in a form that can be handled both by the communication equipment 
which transmits the data and by the computational equipment which 
processes it. U.S. Steel selected typewriters, at the point of origin, that 
produce five-channel punched-paper tape. Five-channel paper tape 
can be fed directly into teletypewriters. Data originally recorded at 
a branch plant or sales office, therefore, can be sent to the central office 
without retyping. At the receiving end a receiving teletypewriter pre- 
pares a five-channel paper tape automatically. The new tape is used to 
produce a report on another typewriter, or is sent to the computing sec- 
tion, where a special machine can produce punched cards or transfer 
the data to magnetic tapes. Figure 6 illustrates such a system for 
integrating sales-order procedure. 

Conversion to Integrated Data Handling. Experience has shown 
that in order to establish an integrated system the old data system usu- 
ally must be changed. The conversion requires many modifications to 
existing forms and procedures. U.S. Steel, for example, completely 
redesigned many of its forms. 

The new systems also require that it be possible to have exact repro- 
duction of forms, as transmitted at the point of origin. In some com- 
panies various sales offices or branch plants now use somewhat differ- 
ent forms and reports. These must be made uniform. 

Also, communication by teletype or leased phone lines is expensive. 



Integrated Business Systems 183 

Every redundant word that can be eliminated will save transmission 
costs. In one company a recent study of business forms resulted in 
reducing the amount of written information by 40 per cent. Substan- 
tial savings resulted from elimination of unessential data, from "whole 
dollar" accounting, and from standardized coding of the data. 



CUSTOMER SALES 
OFFICE 



CENTRAL 
OFFICE 



PLANT 



CUSTOMER 
ORDER 



CLERICAL 

1. REGISTER 

2. ANALYZE 

3. CREDIT 

(OR AT CENTRAL, 



MAIL 



MAIL 



SALES ORDER 



PAPER 
COPIES 

T 

FILE 



TAPE TELETYPE 



SALES ORDERS 



SORTED 



SALESMEN ^_ 
SALES MANAGER 



CONVERT 
I TO 
CARDS 

T 

REPORTS 
TO 



ACCOUNTING 
PRODUCTION 

CONTROL 
SALES 

MANAGEMENT 
FILE 



TAPE TELETYPE. 



SALES ORDERS 
RESORT 



SHIPPING 
DOCUMENTS 
ETC. 



MAIL 



TELETYPE 



TELETYPE 



PRODUCTION 

ORDERS 
INVENTORY 
REQUISITIONS 
ETC. 



PRODUCTION 
STATUS REPORTS 



SHIPPING NOTICES 
— \ INVOICES 



Figure 6 



When data are to be received and transcribed on common media, 
other procedures usually must be changed. The tape must be filed. 
Checks must be provided to ensure that transmitted data will be accu- 
rate, and that data do not get lost, or are not held up unduly, etc. 

Integrated codes are required. Sales orders, package identification, 
shipping tags, production orders, and sales invoices must be coded in 
a manner which provides for detailed identification and checking. 
The code should be designed so as to facilitate sorting by the sales de- 
partment, the accounting department, and so on. Most companies 
have approached this problem by coding one procedure at a time. 
Installation of codes which will facilitate integrated data processing 



184 Electronic Computers and Management Control 

is a major project. So far it has been given only theoretical attention 
by most companies. 

In preparing for a computer, it has been found that special conditions 
in the data should be assigned definite codes so that their use by the 
computer will be efficient. Records ordinarily reflect a great variety of 
conditions. The computer must be warned, by means of the code, 
which condition is present before it can select the proper procedure. 

If the system designer is not careful at this point, he can force the 
computer to make lengthy searches before it takes a useful step. Even 
at best there will be many occasions when the electronic system spends 
more time finding out what ought to be done than actually doing it. 

Economic Results of Conversion. Initial applications of partly inte- 
grated systems have had two important results: reduction in clerical 
labor, and reduction in time of processing data. 

The investment in equipment and in changing the systems has been 
substantial. However, systems studies have paid for themselves. In 
one company, integration of sales orders, production planning, and in- 
ventory control already has resulted in an annual savings of over a 
million dollars. Equipment ( not electronic ) was repaid in four months. 
Before this could be achieved, more than two years of study was re- 
quired. The time was necessary in order to evaluate the needs, pro- 
gram the system and procedure improvements, establish an integrated 
code, redesign the form and reports, and to select and install appropri- 
ate equipment. 

Integration with Electronics. Some companies have adopted a pro- 
gram of gradual conversion to electronics. They believe that they can 
integrate further developments in electronic systems, or automation, 
with their present methods. However, before their electronic systems 
can make full use of common media, such as paper tape, they often 
require special devices, such as paper tape to magnetic tape, or paper 
tape to card converters. 

Some transmitting devices currently under development may not re- 
quire tapes. They can be actuated by other electronic or optical means. 
They may operate much more efficiently than present systems. For this 
reason, even the most advanced installations try to remain flexible. 

Electronic systems can perform operations not possible with mechan- 
ical equipment. (An example is logical comparison— "branching/') 
Systems design can take full advantage of these new abilities. Some 
systems work being done for nonelectronic equipment may be adapt- 



Integrated Business Systems 185 

able for electronic systems, but because of the added capabilities it 
may be more economical to go directly into electronics. 

Any use of equipment requires some systems analysis. Redesign of 
forms and reports, and testing of changes before their adoption are 
needed to assure continuity of operations. Management that has made 
a substantial investment in a mechanical system may be reluctant to 
consider electronics, despite demonstrable advantages, until competi- 
tors have moved far ahead. 

Specific Considerations for Integrated Data Handling. In designing 
an integrated data-processing system it is necessary to start with spe- 
cific recognition of the various needs for the data: the requirements of 
such groups as the outside auditors of the company, the stockholders, 
the government (for tax and other reports), and trade associations. 
Occasional needs, as for newspaper publicity or use in advertisements, 
should also be considered. 

Too often the approach taken at present is to create a data-processing 
system which meets only the inescapable needs, such as for billing of 
customers, preparation of tax returns, and for publishing annual reports. 
To this may be added some kind of cost system, in many cases operated 
by an entirely separate division of the controller's department. Then, 
in progressive firms, a further modification may be introduced by the 
creation of staff analysis groups, empowered to gather and present 
statistics for special studies. Each operating department creates its 
own requirements and may build up its own data system. 

Even though they have a "whole system concept" as the basis for 
their work, systems groups have found it advisable to begin their inte- 
grated studies by working on fairly independent segments of the over- 
all system. They start by attempting to solve the needs of all the inter- 
ested end-users of the data in a particular area of operations. 

In so doing, there is always a danger that practice and theory will 
differ. If the installation is to keep from too narrow an approach, some 
method must be found to remind the individual analysts of the over-all 
objectives of their work. 

In general, the following considerations are typical of those to be 
remembered: 

1. Management's need for planning and control information. 

a. Different types of information for different managerial levels 
( condensed versus detailed data, charts versus figures, extent of 
written explanations, etc.). 



186 Electronic Computers and Management Control 

b. Different frequency of reporting for different purposes. 

c. Different forms of analysis for different purposes (exceptions, 
variances, trends, etc.). 

d. Availability of stored information, to answer special inquiries 
for analysis to prepare special reports, etc. 

2. Routine operating needs. 

a. Daily and periodic accounting requirements, for billing, inven- 
tory, cost accounting, payroll, payables, cash receipts and dis- 
bursements, etc. 

b. Production requirements for scheduling, requisitions, orders, 
etc. 

c. Personnel information for manpower assignments, turnover, etc. 

d. Stockholder information. 

e. Sales and market information. 
/. Capital asset control. 

3. Requirements for internal audit. 

4. Legal and tax requirements. 

5. Requirements for external audit. 

6. Requirements for published reports. 

7. Data for trade association and government agency requests. 

These are only in the nature of topic headings. Under each, a set 
of fairly detailed requirements should be established. 

Measuring the Requirements for an Integrated Data-handling Sys- 
tem. It is not easy to set up rules or guides which can be used to 
choose the best system. The subject has received considerable atten- 
tion in recent years. Systems and procedures groups are to be found 
in almost all large organizations. But in spite of the elaborate manuals 
which they have developed, few are able to define and measure all the 
requirements for integrated data-handling systems. 

An illustration of the difficulty can be seen in the present problems 
of many companies which manufacture and install electronic systems. 
Their engineers have learned how to make machines which will operate 
at startling speeds. But still, they and their customers have had diffi- 
culties in making efficient installations. The manufacturer must rely 
on the customer to detail his needs, but the customer has difficulty in 
doing this in a useful manner. 

There are certain requirements for integrated data-handling systems 
which apply to most, if not all, data-handling routines. These require- 
ments can be described precisely and measured quantitatively. Some 



Integrated Business Systems 187 

consulting firms refer to this as "information engineering." Specifica- 
tions can be established for a communications system just as engineer- 
ing specifications are determined for a building. 

The first step is to define as precisely as possible the various aspects 
of the availability of information and of the need for information. 
Next, the volume of data and rate of flow must be measured. Then 
the communication and processing links can be built. 

This means the study will determine the points of use of information, 
the types of information required at each point, the time when the 
information is needed, and the forms in which it is desired. These 
items are then examined to determine overlaps. In this way unneces- 
sary duplication is eliminated and provision of data for several similar 
needs can be unified. 

Company analysts can determine where original information can best 
be obtained in order to meet the requirements. The type, form, and 
time of record are important. The degree of accuracy required in the 
data is of economic significance, as is the ability of one recording point 
to serve several needs. 

With these items defined, the next step is to measure the volume of 
information which must be handled at both the point of origination 
and the point of use. The rates of flow at various times is also signifi- 
cant. 

Only then can the processing specifications be "engineered." The 
needs of the integrated system then can be related quantitatively to the 
abilities of various devices for data recording, processing, storing, and 
output. While electronic systems offer considerable advantages in 
many situations, all other possibilities should also be considered. In 
some cases, for example, direct observation by the user may still be 
best. 

Processing requires consideration not only of arithmetic calculation, 
but also of "logical" choice, collating, sorting, and of other abilities. 
In addition, storage volume and speed of access are important . 

After the measurable factors have been determined, then other re- 
quirements which are more qualitative can be taken into account. 
Some of the qualitative aspects which should be considered in inte- 
grated data-handling systems are: 

1. A record must be made of all accountable transactions and events. 
In some cases the record must be precise ( particularly for trans- 



188 Electronic Computers and Management Control 

actions involving outsiders ) , and in other cases estimates will do. 
The degree of necessary precision should be determined. 

For some items the source record should be preserved, in other 
cases it is less necessary; the length of time for each should be 
established. In some cases the point of origin of the information 
is vital and should be recorded; in other cases it is not necessary. 
( For example, as a rule it is considerably more important to know 
at what place a customer makes a purchase from the company 
than it is to know at what place she pays her bill. ) 

2. There should be a way for authorizations to be recorded. 

In most cases today authorization is indicated by signature of a 
responsible party, placed on a document. If the proposed elec- 
tronic system bypasses the manager, or eliminates the paper docu- 
ment, some other means of control— perhaps statistical analysis 
with rapid feedback— will have to be established. 

3. There should be a means for subsequent checking. 

The validity of the entries, the accuracy of the processing, the 
propriety of the classifications used in summarizing, and the phys- 
ical existence of items reported on, such as inventory, all should 
be subject to audit. 

It appears likely that in many cases this check will not be an 
actual retracing of data flows, as in some present auditing proce- 
dures, so much as a testing of equipment and systems design so 
as to indicate a high probability that it is performing satisfactorily. 

These principles can be applied to any proper system. They always 
have been considered, to some degree, when installing and monitoring 
a data-processing system. But in the past, each point could be assessed 
as the system analyst went along, as he worked on a particular routine, 
since his decisions would have only a minor effect on other routines. 

The volume or rate of flow of the data did not have to be measured 
too precisely, since the systems usually could be expanded by adding 
more clerks and machines. Now it will be important for all of these 
factors to be measured in advance so that the systems engineer can 
determine the extent to which each must be taken into account in 
designing the system. 

Some work has been done in this area by large companies, but the 
results are not generally available as yet. However, even if they were, 
it appears unlikely that a company could learn more than certain 



Integrated Business Systems 189 

general principles and certain patterns for setting procedures. Spe- 
cific measurements obtained by one company are not likely to be appli- 
cable to the system of another. 

Summary — Integrated Data-handling Systems. Current application 
studies for electronic systems have shown that regardless of cost, size, 
or character, a major requisite is the measurement of the quantita- 
tive requirements. Availability of such specifications is rarely found. 
Therefore studies must be undertaken to gather the data. 

The requirements are generally shown in terms of flow charts or 
process charts, by specific functions. However, some work has been 
done to apply mathematical techniques. These are used to set forth 
the requirements in an orderly and logical fashion, so that the data can 
then be evaluated. Mathematical models already have been built for 
applications such as production scheduling, inventory, control, and 
budgeting. 

The desirability of mathematical model building for data-handling 
systems is becoming recognized. The use of such models has shown 
the following results: 

1. The mathematical models have permitted companies to experi- 
ment with their systems before altering actual procedures. 

2. The models have been helpful in establishing the "engineering" 
specifications for the electronic system. 

3. Analysis of the models has made it possible to eliminate useless, 
outmoded, or redundant information and reports. 

So far, systems analysis has been aimed primarily at improving the 
basic patterns for the "combining" form of flow-data handling. For 
the most part this is the basic structure already employed in data- 
processing systems. Individual items are grouped and regrouped. 
They go into departmental, product, and other cost totals or revenue 
classifications. They end up summarized on operating reports and 
company-wide statements. 

An example of a mathematical method for providing a "descriptive 
model" of the business situation, applicable to a study of recording 
data and the document flow through a present business data-handling 
system, is found in Appendix 4. 



CHAPTER TEN 



Automation and Scientific Computation 



Mathematics and electronics will have far-reaching effects on busi- 
ness operations through their impact on information-processing systems, 
on automation of the office, and on management planning and control. 
But electronics and mathematics also have important applications in 
other major areas— automation of production facilities, simulation of 
problems, and scientific computation for engineering and experimental 
purposes. This chapter will attempt to describe these subjects suffi- 
ciently so as to indicate the significant relationships between them and 
the topics discussed in earlier chapters. 

AUTOMATION 

For years production men have been seeking ways to make their 
operations more mechanical and less dependent upon the human 
worker. In some cases, such as in soft drink bottling plants, this mech- 
anization has been achieved to a very high degree. But though "cam 
and gear" mechanization can be included as part of the significance of 
the term "automation," it is not all of the picture. In most automation 
the machinery does not simply go through a repetitive operation. 
Rather, it is designed to perform any of several operations, dependent 
upon automatically supplied instructions. 

In more complicated situations the machinery can adjust its perform- 
ance, as it operates, in order to keep the product within specified toler- 
ance limits. The machine achieves this control through a feedback 
loop, often called a "servomechanism." 

The "servo" measures the operation as it proceeds, and initiates ad- 
justments to keep the process under control. A thermostat for the 
furnace is an example of a servo. 

190 



Automation and Scientific Computation 191 

The next step is to link a number of machines. Each has its own 
controls, but they are integrated so that the machines will process and 
pass the product from one to the next, in effect as though they were 
part of a larger machine. The various parts perform the necessary 
operations in accordance with a program of instructions. In addition, 
the operations are modified by controls which "watch" conditions as 
they develop. In such a system, various products can pass from one 
end of a line to the other, untouched by human hands. 

The final step in automation, not yet achieved, will be to link the 
production system with an electronic data-processing system and the 
management planning-and-control system. Then all the necessary 
production information would be gathered automatically as the line 
moved, all the information necessary to program, schedule, and control 
operations would be automatically prepared, and the control data 
would be returned from the central information system to the line, as 
required. 

Some experts have compared this system with the human framework: 
the control motors, hydraulic mechanisms, pneumatic valves, etc., are 
the "muscles," the electronic monitors and networks are the "nerves," 
and the central computer is the "brain." (The analogy is useful only 
for emphasizing the integrated nature of the system; as was indicated 
earlier, human mechanisms are so much more efficient than present 
mechanical systems that the comparison is somewhat dangerous.) 

The goal for management is complete integration of the whole sys- 
tem—management purpose, executive decision, production operation, 
marketing operation, and control information. To reach this final step 
it will be necessary to design both the "automation" system and the 
system of planning and control as one entity. The business organiza- 
tion must be balanced, with the various activities mutually supporting, 
their purposes consistent with the purpose of the organization as a 
whole, in the same sense that the members of a healthy human are 
so coordinated. 

Requirements for Automation. It has been found that the require- 
ments for automation have much in common with the requirements for 
systems of electronic data processing. Innovations have not been 
easily introduced. Most installations have been of devices for handling 
specific parts of the whole operation. These installations have reduced 
manpower needs, increased productivity, and lowered costs. However, 
as with data processing, the full benefits of automation are expected to 



192 Electronic Computers and Management Control 

be achieved only when the devices can handle the whole process— can 
start with the inputs, raw materials, and carry through to the outputs, 
the packaging and delivery of the product. 

As with the development of electronic systems, the government has 
been a primary source of support for research and for pilot installations 
of automatic controls. Weapons research, both for control of the 
weapons themselves in combat and for automation of their production, 
has been a major project of the Defense Department in recent years. 
In addition to encouraging manufacturers by bearing the costs of this 
research and development, the government has sponsored research di- 
rectly by supporting a number of groups located in universities and 
other research centers in various parts of the country. Unfortunately 
much of the work of these groups is still classified as secret. However, 
some of their findings have been made available to business, and it can 
be hoped that the flow of information will increase as the reasons for 
classification dwindle with time. 

System Analysis. Automation, like integrated business-data systems, 
requires that the production plant be studied as a whole. It has been 
found necessary to consider the product, its design, the components 
which compose the product, its functional use, and so on. The re- 
quirements must be analyzed to determine what can be combined, 
simplified, or eliminated, using automatic equipment. The whole sys- 
tem must be geared to its purpose, in the same sense that the design 
of a bridge or the architect's plans for a building are based on the 
purpose of the structure. 

Attempts to introduce automation have uncovered the need for de- 
velopment of a body of knowledge to bridge the gap between automa- 
tion engineering and managerial planning and control. Otherwise, for 
example, automation has sometimes solved the production problem so 
well that it created new financial or sales problems. Poorly planned 
automation has unbalanced the production line when applied only to 
a part of the production system. Unless applied to the most serious 
problems, the more successful it is, the worse bottlenecks it creates. 
Automation must be programmed so that it is applied at the right 
places and in the right order of application. 

Personnel Change. Automation usually has resulted in substantial 
reduction in the labor force and in needs for middle management, 
with all their supporting functions. Many problems of change-over 



Automation and Scientific Computation 193 

have had to be met. Besides the rather obvious changes, such ac- 
tivities as communication have had to be reprogrammed. Consider- 
able information which was accumulated, transferred, processed, and 
analyzed as part of the duties of the displaced personnel now is being 
handled automatically. Devices are needed to monitor the automatic 
machines as well as to provide the information which is needed for 
planning and control. Sometimes new kinds of information have to 
be provided. 

Compatible Design. Centralized automatic controls, while avail- 
able for single situations, have not been developed for larger, complex 
installations. Design of automatic, centralized control equipment will 
require that an over-all framework be established for the factory, 
much like that for electronic systems in management planning and 
control. The various functions must be carefully defined as to their 
requirements, accuracies, and speed of operation. Equipment at each 
substation must be sufficiently reliable to correspond with the needs 
of, and be compatible with, the operation immediately before it and 
after it. 

Illustrations of Compatible Design. The Ford engine plant em- 
ploying some eight thousand people at Dearborn, Michigan, is an 
example of partial application of automation. It uses automation 
equipment to do the heavy work of lifting, turning, and moving bulky 
engine blocks and other parts through machining, assembling, and 
testing. 

Another Ford installation is an example of more complete automation 
of a process. A system about the size of a football field performs 540 
separate operations to turn out engine blocks at the rate of 100 an hour. 
A few men at control boards monitor the whole system. 

In the field of distribution, Lever Brothers have a system which re- 
ceives cartons of various products, intermingled. It automatically 
sorts, shunts them onto individual feeder lines, loads them on pallets, 
and delivers them to docks or to the warehouse floor. Sorting is han- 
dled by electric eyes, which read a series of identity stripes on the 
cartons. The eyes actuate switches according to the dimensions of the 
stripes. 

Two of the leaders in the production of equipment for automation 
are General Mills and United Shoe Machinery. Both companies have 
been so successful in developing devices for their own use that as a 



194 Electronic Computers and Management Control 

result they have found new products. The markets for these products 

are far different from those in which the companies were originally 

engaged. 

As might be expected, all manufacturers of electronic equipment are 
strongly interested in automation. Admiral is an example of a company 
far advanced in this field. 

Flexible Production. An important aspect of automation, not al- 
ways fully recognized, is that the products of such a system need not 
be uniform. It is possible to govern the controls on machinery by 
means of systems using magnetic tape or punched cards. By using 
a computer to prepare these "instructions," it is possible to obtain a 
different set of specifications for each product that comes off the line, 
if desired. 

Or, as appears more likely, automatic equipment can be set to assem- 
ble a variety of parts. Each of the parts can be mass-produced, but 
the combinations can be varied. Automobile assembly lines already 
achieve this with various color combinations. 

Automation of the office follows similar principles. Certain elements 
of each transaction record are unique. These can be stored for refer- 
ence, but they are screened out in the processing. The common ele- 
ments are grouped in various ways on different reports. This "com- 
bining" can be handled automatically in an electronic data-processing 
system. 

Automation — Limitations and Obstacles. Automation is introduced 
only where it will pay. Most physical operations in business can be 
performed by a machine. The limitations are largely economic. It is 
necessary to determine whether it is worthwhile to spend the money to 
design, build, and operate the necessary equipment, especially in view 
of the research that must be done to fill some of the present gaps in our 
knowledge. 

Technological Obstacles. The rule in the past has been that the 
simpler, more repetitive, and more frequent the operation, the more 
likely that it can be successfully mechanized. But with automation 
a new element is introduced. It is no longer necessary that the proc- 
ess be exactly repetitive. Instead, the requirement is that the pur- 
pose of the process be known and measurable and that the operations 
be controllable. As in programming and scheduling of management 
decisions, this places considerable emphasis on the control-feedback 
mechanisms. 



Automation and Scientific Computation 195 

The significance of this can be seen in many current installations, 
particularly in modern petroleum refineries. In typical systems there 
is a combination of completely automatic minor controls, combined 
with over-all human control by operators who watch instrument panels. 

In general, automatic control can be established wherever the rela- 
tionship between the incoming information and the required correction 
can be predicted. Such measurements as those involving pressures, 
temperatures, rates of flow, and levels are frequently of this type. 

Where the process is more complex, measurement of all the variables 
is seldom accurate enough to permit completely automatic control of 
everything. There are points at which it is more economical to rely on 
the human operator for control. An intelligent human is a very effec- 
tive "device" for this sort of operation— he has rapid and reliable input 
( reading dials, etc. ) , rapid access to a large storage ( his memory ) , and 
efficient output (pressing the right buttons, etc.). 

Most important, human judgment can compensate for feedback in- 
adequacies. The essence of human control is the fact that an intelli- 
gent man has the ability to control without complete information. Of 
course, the better the information presented to the human, the more 
effectively he can operate. Control panels in refineries, etc., are built 
to give a remarkably good picture of the process flow, but even they 
are not sufficient to make the process control foolproof, without human 
monitoring. 

Personnel Limitations. Automation requires personnel trained in 
systems engineering. They must be able to integrate the various 
functions and subfunctions of a business. Few such engineers are 
available today. Most engineering schools have encouraged speciali- 
zation, such as mechanical engineering, electrical engineering, civil 
engineering, and so on. 

In addition, the mathematical training usually given engineers, 
while greater than that in most other fields, is frequently not suitable 
or sufficient for systems engineering. As a result, most engineers are 
capable of automatizing only separate elements of a process, not the 
whole. 

Few good systems engineers are at present available to industry, 
compared with the demand. The advance in automatic military sys- 
tems has led the government to train some systems engineers, but the 
number of engineers is not large and their training is not always suit- 
able for commercial work. 



196 Electronic Computers and Management Control 

High Cost. Automation usually requires a substantial investment. 
New machines are required, as well as better instrumentation and 
more accurate and flexible controls. Salaries for competent personnel 
are high, and their work takes time. Systems engineers need the help 
of mathematicians and of production engineers for defining the control 
problems, for measuring the relevant variables, and for designing the 
integrated system. 

Project Tinkertoy, which was sponsored by the Navy to produce 
electronic circuits automatically, required almost $5,000,000 to de- 
velop a pilot plant. 

Eventually, the cost of automation should be substantially reduced 
through improvements in production machines, in the logical design 
of computers, and by other means. Meanwhile, partial applications 
will continue to be fairly expensive, though in particular cases they 
often can rapidly repay the necessary investment. 

Summary. Fully automatic factories will be built some day, but for 
most industries that day is not tomorrow. Devices are now available 
for partial applications, particularly for material handling and for con- 
trol of certain types of machine tools. It is this type of automation 
which is being attempted first. 

The requirements of the Department of Defense will continue to 
encourage development of automatic weapons, and automatic means 
for their production. Use of the knowledge gained in this research may 
be expected of many businesses which seek government orders. 

Successful automation by competitors is a threat that businessmen 
cannot afford to ignore, even though they might prefer to wait until 
better systems become available. Despite many obstacles, accelerated 
use of automation appears inevitable. Because of these pressures, the 
problems of the executive will be increased over those that might be 
expected if development of automation could be studied more leisurely. 

SCIENTIFIC COMPUTATION, MATHEMATICAL COMPUTERS 

Still another major development which must be considered is the 
availability of computers designed to do scientific computation. These 
were the first models of electronic computers. Complicated engineer- 
ing problems almost always require a certain amount of mathematical 
computation, and advanced scientific work relies heavily on mathe- 



Automation and Scientific Computation 197 

matics. As a result of these needs, and spurred by the large sums of 
money available for defense effort, the development of electronic equip- 
ment for high-speed computation has proceeded at a remarkable rate. 
Most of the equipment now offered for data processing has descended 
directly from these scientific computers. In some cases all that has 
been done to change over the design has been to improve the input 
and output capacities, and in some cases the storage. 

Electronic computers designed primarily for computation already are 
used by business to a surprising extent. Volumewise, it is probable 
that more computation has been carried on by them in the few years 
since 1949 than by all other means in all prior human history. 

The cost of electronic data-processing systems is so substantial that 
it will be easier for executives to justify acquisition if other uses, such 
as for engineering or actuarial computations, can be shown in addition 
to the routine data processing. On the other hand, it has been found in 
many cases that if needs for large amounts of computation occur irreg- 
ularly, they can be served better by making use of the computing 
services established in various parts of the country. Most scientific 
computers are available on a rental basis, by the hour. Also, manu- 
facturers of business computers are already offering this type of service. 

For example, a number of Detroit manufacturers cooperated to pur- 
chase a Burroughs computer, now installed at Wayne University. 
Among others, General Motors and Ford contributed substantial sums 
to this installation. The money is being used for studies of production 
scheduling, engineering computations, and for market analysis. 

These centers also will handle what is coming to be known as "data 
reduction," the routine processing of large amounts of data, such as 
observations made during a scientific experiment, or regular production 
reports. Data reduction may make use of complicated mathematical 
formulas, but it is not experimental, in the sense that the formulas used 
are known and tested. The operation may be compared with assembly- 
line production of physical objects. For example, General Electric has 
found that there is so much work of this sort for its IBM 701 at Cin- 
cinnati that it has arranged to send extra work (beyond a 24-hour-a-day 
load) by wire to other machines, including the IBM service center in 
New York. The reader will recall that the 701 is a machine which can 
make several thousand computations a second. 

Data reduction is coming to be a sizable business by itself. A num- 



198 Electronic Computers and Management Control 

ber of firms have been set up to handle such work on a contract basis, 
and university computers receive considerable support from this source. 
The next section will describe briefly some of these activities. 

Some Problems Solved by Scientific Computers. It is difficult to 
generalize as to types of problems which have been solved by these 
mathematical computers. Almost every scientific field has been af- 
fected. However, description of a few problems, together with men- 
tion of others, may show how these machines have been used. 

Many types of problems involve the use of prepared tables. The 
machine can store these tables in its tape or high-speed memory, and 
refer to them at appropriate points in the computation. The tables are 
first prepared by the machine itself, of course, following predetermined 
formulas. In many cases it is not necessary to prepare a whole table, 
since the machine can also recompute from the formula as it goes along, 
unless the formula is so complicated that it fills the high-speed storage 
of the machine. 

Many problems are of a type in which the result of one computation 
is used in the next computation. The machine can carry out millions 
of computations, in which this "retracing" occurs in the sense that the 
machine does the same sort of calculation over and over, using the 
result of the last computation as a factor in the next. Thus the same 
set of instructions will serve for a major part of the "program." 

For example, to calculate the path of a projectile, the machine can 
weigh such factors (mathematically speaking) as the weight of the 
object, the speed of acceleration, the force of gravity, the air resistance, 
the curvature of the earth, and so on. It calculates where the projectile 
will be a split second after firing. At this moment the projectile will 
have a certain position and speed, which the machine uses in calcula- 
tion of the position and speed a split second later, and so on. The 
result can appear either as a table of figures or as a plotted series of 
positions on graph paper. 

Carrying the same example a step further, the computer can operate 
in "real time," while the projectile is still in the air. Through use of 
telemetering devices, information concerning the flight is sent back to 
the computer. The computer produces answers in a form which can 
be sent back to the projectile in order to control its course. 

In other cases, the computer may be part of the projectile. The com- 
puter may evaluate such matters as the relative positions of the 
projectile and of the target which the projectile is approaching. The 



Automation and Scientific Computation 199 

computer system converts radar readings to mathematical form and 
correlates them with the position of the projectile and the controls on 
the projectile. This type of electronic control loop is the basis for 
much of our country's guided-missile effort. In a real sense the elec- 
tronic computer is part of our first line of defense. 

The use of a computer for such purposes may at first glance appear 
to be a far cry from any commercial application. However, knowledge 
of these developments has led to such statements as that by an execu- 
tive of General Motors, to the effect that they could now build a car 
which would steer itself down the highway. The executive quickly 
pointed out that it would be uneconomical though technically possible 
to do so. 

Whereas it may well be questioned whether it will ever be desirable 
to produce a self-steering car, the businessman must be alert to the 
possibility that some such application of computers may become of 
great importance to his product. An example of a system which comes 
close to being of this order is the elevator dispatch control now being 
installed in large office buildings. This electronic system gathers infor- 
mation as to the position of the cars and as to the buttons pressed to 
demand service; it then computes an optimum car allocation and dis- 
patches the cars automatically. 

Differential Equations. Numerous problems which confront the 
engineer involve measurements which give rise to relationships that 
are expressed in differential equations. For example, the distribution 
of magnetic-flux density in an electromagnetic device, or the distribu- 
tion of stress in an elastic body, or velocity and pressure in a fluid are 
measured with "partial differential equations." 

A major difficulty with these situations is that to solve these equations 
usually calls for a carrying out of a great number of computational 
steps. To handle these problems it has been found economical to 
design and build special types of computers which analyze differential 
equations. Both analog and digital systems are used for this purpose. 

Success with these special-purpose machines has led researchers to 
hope that similar special computers, designed to incorporate improved 
mathematical methods, may some day enable them to handle a variety 
of business problems which are now beyond reach because of the com- 
plexity of the situations involved. 

For example, the possible routes in a production shop, with attendant 
costs and times of completion, give rise to equations of fantastic com- 



200 Electronic Computers and Management Control 

plexity, using present knowledge. It is believed that some new variety 
of combinatorial mathematics, together with new computer designs, 
will be required if such problems are to be handled electronically. 

Another problem is the representation of minimum cost and maxi- 
mum revenue curves. Quadratic equations, or even more complicated 
models, are required to produce curves that "peak" and "valley" in this 
fashion. Again a special type of computational design may someday 
give a technological breakthrough in this area. 

EXAMPLES OF COMPUTATIONAL APPLICATIONS 

To round out the picture of the scientific computer and of large-scale 
computational problems, a few applications will be described, followed 
by mention of other problems on which these machines have worked. 

The RAYDAC and Teleplotter at Point Mugu. The Raytheon Digi- 
tal Automatic Computer ( RAYDAC ) was developed for the U.S. Navy 
Bureau of Aeronautics to analyze guided-missile flight problems. 

When a guided missile is launched, ground radar stations record its 
speed, altitude, and direction. Inside the missile miniature transmitters 
send out information on the speed, how fast it is using up its fuel, the 
temperature of the missile's engine, its angle of ascent, and so on. All 
this is coded and given to the computer, where the problems of flight 
are analyzed to determine such matters as stability of the design, tend- 
ency to yaw, etc. 

The data are punched on cards, which are given to a "teleplotter." 
This device is a machine about the size of an ordinary office desk. It 
can translate numerical data into chart form, through use of a swinging 
head that counts the grid lines on a chart photoelectrically, and auto- 
matically plots positions as they are fed to the machine. 

Incidentally, given data such as are customarily developed for a 
break-even chart, this device could draw the chart in a few minutes. 
However, the machine is able to work with much more complicated 
data than are usually developed for this purpose by business analysts. 

The value of this equipment lies primarily in the speed with which it 
makes the results of an experiment available to the researchers. In the 
past it was necessary to gather data at a test station, then carry them 
back to offices for days or weeks of computation to discover what hap- 
pened. Now it is possible to analyze each flight as it occurs. Analo- 



Automation and Scientific Computation 201 

gous advantages for various types of business experimentation are 
obvious. 

UNIVAC and the Election. One of the most publicized statistical 
uses of electronic computers occurred on election days in November, 
1952, and November, 1954. In 1952, for example, the UNIVAC was 
used to make a forecast of the presidential result at a time when less 
than 6 per cent of the vote was in, with no returns at all from 21 states. 

The situation is worth some discussion, since it illustrates the careful 
planning which must precede the handling of data on the computer. 
About a month of working time was spent to formulate a theory of 
election mathematics, which detailed equations showing the relation- 
ships between early returns and the final results in previous elections. 

The preparation for the application lagged, and as the deadline ap- 
proached more and more workers were added to the project. The 
final "program" was completed on election day. There was no time to 
test it out. 

When first returns of about 3,400,000 votes had been received, they 
were fitted into the equations, given to the machine, and the calculation 
started. The actual machine running time was about 6 minutes, after 
which a detailed forecast of the probable results in each of the 48 states 
was printed out. The forecast gave 5 states to Stevenson and 43 to 
Eisenhower. Since the program had not been tested, the group de- 
cided that this might be erroneous and withheld the first prediction. 

In the 1954 election the most interesting event was the mistaken 
prediction which resulted from an inversion of the data. At one point 
the machine was given Democratic returns as Republican, and vice 
versa. The machine was not programmed to catch such errors and so 
gave an erroneous prediction. 

In neither case were any of the errors the fault of the machine. This 
case forms a perfect illustration of two possible sources of difficulty in 
using electronic equipment. The first election forecast was not trusted 
because it had not been possible to develop a formula which showed 
the exact relationship between early returns and final results. The ap- 
proximation, however, was surprisingly good. The second was a 
human operating error under hectic conditions. 

The election case is of interest for another reason, of a more theoreti- 
cal nature. For many situations, especially in the physical sciences, a 
researcher attempts to find a formula, or a set of rules for a procedure, 



202 Electronic Computers and Management Control 

which leads to a specific, precise answer to the problem. In business, 
however, such precision is seldom necessary. The situation often is so 
complicated that the analyst may not be justified in trying to find a 
formula that will precisely relate all the variables. Instead, a numerical 
approximation may be sufficient. If this shows only some statistical 
probability, it may provide a useful solution to the business problem. 

The election forecast was of this type. An examination of past rela- 
tionships showed that when one group voted in a certain way, then 
another group had voted in a certain way. This was expressed as a 
numarical relationship. There was no need for the forecasters to worry 
about why this had been so, unless there was some reason to believe 
that events had changed in such a way that the old relationship might 
no longer hold. ( An example of this might be the case of a native son 
candidate who might be expected to carry his home state with a higher 
presidential vote than his party would otherwise poll. ) 

Given a set of these numerical relationships, the machine showed 
what the result probably would be. Such an indication often is all that 
is required as a basis for an operating decision. 

UNIVAC at the Bureau of Census. The first UNIVAC was installed 
at the Bureau of Census in March, 1951. By employing the first large- 
scale computer designed to perform accounting and statistical opera- 
tions, the Bureau of Census has rightfully claimed to be following in the 
footsteps of the Bureau in the last century, since the first large-scale 
application of punched cards was made for the calculation of the 1890 
census. 

The UNIVAC was assigned part of the preparation of the "Second 
Series" Population Tables, which contain 30 tables covering age, sex, 
race, country of birth, education, occupation, employment, and income. 
The tables show statistics for thousands of categories— for each county, 
city, rural farm, and rural nonfarm area within a county. 

The raw data were prepared in the form of a punched card for each 
individual in the United States. The data from the cards were then 
transcribed onto magnetic tapes. The tapes were used to tabulate each 
individual's characteristics, in groups of about 7,000. Then the groups 
were used to produce data for the categories, from which the tables 
were prepared. The tables were printed out on the machine's printer. 
The operation was primarily one of addition, to a number of predeter- 
mined totals. 

In reporting the results of the UNIVAC operation the representatives 



Automation and Scientific Computation 203 

of the Bureau have estimated that the cost of this operation was about 
half that of their regular method. They also reported that the machine 
had operated so accurately that they had not found any errors which 
could be traced to machine malfunction. 

Air Force Programming. At an unclassified conference held in 
Washington in 1953, a report was made of an application under way 
to use an electronic computer for an Air Force problem. 

The model problem was designed to provide a document listing of 
the amounts of selected material needed to support a war plan. Avia- 
tion fuel and lubricants, ammunition, guided missiles, aircraft guns, 
assist take-off units, auxiliary fuel tanks, pylons, chaff, photographic 
film, boats, drop kits, and drop sondes are examples of the classes of 
items for which detailed item requirements were computed 

Monthly schedules of requirements were obtained for each of the 
material items after an assumed D-day. Allowance was made for the 
expenditure of material by combat aircraft and by aircraft engaged in 
other missions, such as training. Important kinds of losses, stock levels 
at installations and depots, and transit times entered into the deter- 
mination of requirements. 

It was determined that when data are prepared for the computer, it 
is most efficient to arrange them in a manner which simplifies the hand- 
work and writing of numbers because these operations, when per- 
formed by people, have a greater error rate and are much slower than 
those performed by machines. 

The machine was programmed to make computations which would 
yield levels of material requirements by months. These were printed 
out in tables with descriptive headings. 

In making the computations the machine used a reel of tape which 
contained the data from the war plan. These data dealt with flying of 
diverse kinds. Schedules of deployed aircraft by type, model, and 
series were shown for each month, for the continental United States 
and for overseas. The number of various kinds of aircraft were listed 
for the various kinds of schooling and practice, such as crew training, 
joint training with the Army, tactical support, individual training, ad- 
ministrative flying, testing, and special mission. Information on crews 
and on individuals completing training each month were also listed. 
In addition, there were schedules of aircraft accepted by the Air Force 
from manufacturers for delivery, and allocations to fighter-bomber 
interceptor units and other kinds of units. 



204 Electronic Computers and Management Control 

On another reel were the data which showed monthly hours of use 
for each kind of aircraft, gallons of fuel consumed (per flying hour), 
material items expended per sortie, rounds of fire in life of weapons, 
material loss factors, and pipeline times for transit and for installation, 
etc. 

Three blank tapes were mounted in the computer to handle the re- 
quirements of computation and results. The results were then rear- 
ranged, edited, and merged with descriptive alphabetic information, 
before printing. 

The instruction tape in this problem contained a central computing 
routine 650 words long, with 12 characters to a word. On a punched- 
card run of a similar problem, such a computation routine previously 
took some 16 plugboards containing 3,818 wires. Preparing the com- 
puter instructions took about as long as wiring the plugboards. How- 
ever, once prepared, the computer instructions could be stored in a 
sewing thimble, while the plugboards took several cabinets. Moreover, 
the machine automatically selected the next routine from the tape, 
while humans had to be entrusted to select the right plugboard. 

The program data used consisted of some 15,400 activity levels. The 
computed results contained 57,500 activity levels. The planning fac- 
tors used numbered about 144,000. 

In addition to this problem dealing with combat consumables, the 
Air Force has experimented with test computations of military person- 
nel requirements by Air Force specialty code, aircraft-engine procure- 
ment, overhaul and shipping requirements. 

An interesting result of the experience was the discovery that the 
greatest difficulties were encountered at the start and at the end of the 
experiment. The experimenters were faced by the fact that many of 
the decisions they were measuring had always been made on a judg- 
ment basis, were not governed by the same factors as previously used, 
and thus were difficult to estimate. 

The conclusions were that this approach promoted consistency in the 
scheduling process, allowed a self-checking computer to do the tire- 
some and time-consuming part of the work, permitted a systematic 
presentation for review, provided for rapid recomputation when pro- 
gram assumptions led to unacceptable results, and released manpower 
to be used on planning factors and on other areas which require judg- 
ment. 



Automation and Scientific Computation 205 

Bureau of Aeronautics. The Bureau of Aeronautics must deter- 
mine when to order government-furnished equipment for airplane- 
production lines. Some items are needed as much as 18 months be- 
fore the airplane is scheduled to be completed, while some items need 
not be furnished until 3 months after completion, when the plane is 
delivered to the point of use. 

One piece of equipment, such as an engine, may be delivered to as 
many as 325 possible destinations— one of the services, a foreign coun- 
try under Mutual Security, the fleet, etc. Quantities required may dif- 
fer for each. A single plane may be furnished with from 25 to 230 
items of aeronautical equipment. 

Changes average about 50 a week in the equipment alone. Revisions 
of schedules and quantities occur frequently. 

About 70,000 punched cards were maintained by a staff of 22, work- 
ing full-time with considerable overtime. When an important change 
in the schedule occurs, as many as 50,000 of the cards may have to be 
changed. 

The problem was placed on a National Cash Register's CRC 107. 
Source data were placed on two master tapes. The first, called the 
airframe tape, consisted of 10,000 words of computer information. 
Each airframe was identified by a particular code stored in the address 
word, followed by space for 50 words of information for that airframe. 
These include the manufacturer and plant location, fiscal year, end 
user, source type, contract number, priority, unit cost, total cost, total 
quantity of airframe for fiscal year, total to date, balance due, month in 
which the schedule first starts and when it ends, and a monthly sched- 
ule of airframes for the next 3 years. 

The other master tape was called the equipment tape. It consisted 
of some 60,000 words of information. Each piece of equipment had a 
code in the address word, followed by the codes of the airframes that 
used this equipment. The fiscal year of use, end user, source type, 
quantity of equipment per airframe, installation lead time, and appli- 
cation were stored as information. 

The procedure started with a particular piece of equipment. The 
computer was programmed to search the airframe tape for the first air- 
frame noted as using this equipment. When a match was made be- 
tween the corresponding airframes on the two tapes, a further search 
was made down the airframe tape, to match the corresponding end 



206 Electronic Computers and Management Control 

users and fiscal years. This search was required since the same air- 
frame was reproduced several times on the airframe tape, for varying 
fiscal years and end users. Once the match desired had been achieved 
at all points, the computation consisted of determining the monthly 
quantities of required items in the form of a schedule— which thus 
became the "program" for operations. 

Sales Analysis. A study conducted by the Retail Department, Bu- 
reau of Advertising, American Newspaper Publishers Association, made 
predictions on how and when people would buy 28 basic lines of mer- 
chandise over a period of a year. Trends as reported for the years 
1941 to 1952 by the Federal Reserve Board sales in department stores, 
nationally and by districts, Department of Commerce seasonal sales, 
trade association reports, and analyses of advertising by 91 commodi- 
ties were correlated by the least-squares method. The UNIVAC was 
used to prepare the tables, which showed that 1954 patterns could be 
expected to parallel closely the 1952 experience. 

Based on this confirmation, the Bureau issued a 58-page "Annual 
Time Table of Retail Opportunities" to its 1,012 member newspapers. 
The study had showed that during the years 1941 to 1952 the seasonal 
pattern of total department-store sales fluctuated less than 2 per cent 
in any one month, despite the war, scare buying, and so on. 

A series of 336 trend lines were calculated in 3% minutes of machine 
time. Comparable hand computation would require over 200 man- 
hours. Programming the machine required considerable time, which 
cut down the margin of saving, of course, but if the formulas hold for 
another year, the program can be reused. 

Summary. There are well over 100 large-scale electronic computers, 
designed and installed by universities and research groups in this coun- 
try and abroad. Almost all of these are engaged in what can be called 
"scientific computation." In addition, hundreds of large and medium- 
size commercially produced computers are working for industry on 
data reduction. 

The use of these machines has opened new horizons for engineers 
and scientists. Many problems which formerly were avoided because 
the computation required was known to be too large to be handled 
economically now are treated on a routine basis. 

General-purpose computers designed for business data processing 
usually will handle scientific computation. Groups studying business 
computer installations can justify part of the cost by demonstrating 



Automation and Scientific Computation 207 

that the systems will aid not only the engineering department, but also 
make economically feasible the development and use of mathematical 
models for production, marketing, and financial analysis. 

SIMULATION 

Simulation is a term which is analogous to model building. The term 
signifies the construction of devices which solve problems by imitating 
the physical characteristics of a situation, or the translation of these 
characteristics into measurements which can be manipulated in a 
mathematical sense. 

The devices are often in the form of an analog system. Sometimes, 
if versatile enough, they are in the form of an analog computer. An 
analog computer can solve many of the same types of problems as do 
digital computers. The difference lies in the fact that the measure- 
ments are not precisely quantified into digital form, except perhaps 
at the input and output of the computer. 

Digital computers also can be used for simulation. 

A simple example of an analog system is the slide rule, in which 
distances simulate the logarithms of numbers. Other mathematical 
relationships can be simulated in physical terms through the use of 
various types of equipment. A device much used in defense work 
employs gears and wheels, which rotate against each other in certain 
ratios, giving quick and reliable answers for such problems as fire 
control for weapons. 

As a rule analog systems are relatively easy to construct, cheap, and 
easy to link to other devices. They are now used in many types of 
situations, varying from controls that are hardly more than thermostats 
all the way up to quite complicated equipment, such as the well-known 
Link trainers that simulate air flight for ground instruction of aircraft 
pilots, or devices that offer military commanders hundreds of choices 
in "war-game" simulators. 

As the devices become more "general-purpose" they approach the 
status of a large-scale computer. Computers can be made to handle 
complicated mathematical transformations to a high degree of pre- 
cision. They have been used to solve complicated aerodynamic prob- 
lems. 

It is also possible to make relatively inexpensive analog devices that 
are accurate to sufficient decimal places so that they can operate on 



208 Electronic Computers and Management Control 

numbers and give digital answers. The extent to which this ability will 
be developed in the future is uncertain, especially for business purposes 
which require digital information. However, it is a field which the 
executive should watch for development of devices which can aid his 
control operations considerably. 

Claude Shannon, who was one of the first to set down the logical 
analysis which underlies much of modern digital computer design, 
also has given an indication of the direction that new improvements 
may take. He and other researchers are studying electronic systems 
that can handle nonnumeric operations. These systems perform logical 
analysis, translate languages, design circuits, play games, coordinate 
sensory impressions, and, generally, assume complicated functions asso- 
ciated with the human brain. 

Nonnumeric computation will be handled by a new type of elec- 
tronic computer, one capable of carrying out a sequence of orders re- 
lating not to operations on numbers, but to physical motions, opera- 
tions with words, equations, incoming sensory data, or almost any 
physical or conceptual entities. " 

Other analog devices make use of such things as mechanical shaft 
positions, signal frequencies, and light intensities. The accuracy and 
reliability of analog systems, while adequate for many purposes, are 
limited by the degree to which it is practical to design and construct 
precision instruments. Digital computers, of course, are completely 
accurate, once the input is established, as long as the system is func- 
tioning properly. 

Analog devices have several advantages. Analog control is usually, 
in essence, a working model of the control function being performed, 
so the design of the system may require only a straightforward con- 
nection of components. Another advantage is the continuous nature 
of an analog device, which often enables it to function rapidly in a 
closed-loop feedback system. 

In a digital system, the computer must sample the data, convert it to 
digital form, transfer it to some storage, call on the stored program, 
compute, transfer data back to the control mechanism, reconvert from 
digital, and finally control. As against this, an analog system may 
require only one electronic conversion, just as the familiar analog con- 
trol of rotating weights on a steam engine requires only a simple lever 
and gear connection. 

A new concept, which may amount to a technological breakthrough, 



Automation and Scientific Computation 209 

is to use discrete physical events to record a "pulse train" which is never 
converted to digital form (i.e., numerical), but instead is handled in 
semidigital, semianalog fashion by an electronic conversion device. 
These techniques will be particularly applicable to automatic control 
problems. The use of these techniques has already reduced the num- 
ber of components, the size, weight, and the cost of certain electronic 
equipment to about a tenth of former requirements. Further reduc- 
tions are considered feasible. 

It is possible, for example, that some sort of analog devices may be 
developed which will simulate business problems in the same way that 
a utility power system is now simulated and controlled through use of 
model grids. 

The Swiss have developed a very complicated mathematical model 
of their govenrment hydroelectric system, and plan to control the 
system by using the model on a computer. The economic system of 
the United States has been represented in model assemblies of elec- 
tronic circuits, a number of which have been built at various univer- 
sities as teaching devices. 

An operating example of an analog system is an incremental cost 
computer used by a Southern utility. Current fuel costs at each 
producing station are recorded and transmitted to the computer. 
Transmission losses and generating costs are statistically determined. 
These generation values are transmitted directly from the stations. 
The computer calculates the cost of delivered power, and the result is 
used to control the system by adjusting the load at each station. 

Conversion Devices. In addition to analog devices, there is growing 
interest in the new field of instrumentation devoted to counting or con- 
version of such conditions as speeds of rotation. Devices convert 
physical conditions into measurements which can be used as input to 
either digital or analog computers. 

The developments are too technical and varied to be discussed here 
in detail, but they form an important part of the electronic picture. 
Through their aid it is becoming possible to measure accurately many 
types of processes as they operate, rather than having to wait to count 
the finished products as they emerge. Obviously this increases the 
possibility of rapid feedback and control. 

A very simple example would be the possibility of measuring traffic 
flow in the aisles of a department store, as an indication of sales 
activity. Many sales managers get a feel for the day's sales merely by 



210 Electronic Computers and Management Control 

observing the condition of traffic in the aisles. Rather simple elec- 
tronic devices can meter this flow and show it on dials or graphs in the 
upstairs executive offices, if there really is value in such information. 
A more elaborate system could keep running totals of amounts rung 
up on cash registers. 

Devices are now being installed in steel mills, and on large-scale 
assembly lines, to monitor production operations. They can give 
quick, reliable information concerning process flows. 

Other Types of Problems Which Have Been Handled by Com- 
puters. Some idea of the immense variety of problems which have 
been handled by electronic systems may be obtained from the follow- 
ing partial listing of such activities. They have been used to: 

1. Compute the flight path of a projectile. 

2. Investigate atmospheric turbulence. 

3. Determine spheroidal wave functions. 

4. Calculate the shape of nuclear-magnetic resonance absorption 
lines. 

5. Study shock waves, a two-dimensional grid of concentrated 
masses subjected to impulsive loads. 

6. Determine the vibrational frequency spectrum of copper crystal. 

7. Do pipe-stress analysis. 

8. Do rotating-disk stress analysis. 

9. Analyze turbojet performance. 

10. Analyze wing flutter. 

11. Do meteorological computations. 

12. Study traffic flow on freeways. 

13. Make geological analysis of oil structure. 

14. Do work measurement. 

15. Prepare tables of the incomplete beta function. 

16. Study stability parameters of aircraft. 

Computers are at work in every major country in the world. Cer- 
tain developments have already come from abroad. Barring increased 
barriers to transfer of information, this international assistance from 
friendly nations should continue. 

On the other hand, the Russians are evidently well along in elec- 
tronic systems design. Some of their scientists have been given high 
recogniticn for their computer work. Estimates have been made that 
show electronic development is a major item in the Russian budget. 



Automation and Scientific Computation 211 

Their magazines discuss automation as though it were to be taken 
for granted. 

The free world has no monopoly on this knowledge. 

Conclusions — Automation and Scientific Computation. The future 
use of electronics and mathematics in business is by no means limited 
to data processing and decision making. In addition to the handling 
of information, electronics will play a large part in the automatic 
handling and processing of materials. As engineering and scientific 
designs become more complicated, more and more use of mathematics 
can be expected together with need for large computation capacity. 

The business executive will find that personnel requirements and 
other such problems tend to be similar in the data-processing, automa- 
tion, and scientific-computation fields. Moreover, integration of the 
system is necessary if optimum benefits are to be achieved. 



CHAPTER ELEVEN 



Role of the Executive in Selection of an 
Electronic System 



The diversity of electronic equipment available and the variety of 
ways in which the equipment can be utilized offer both an opportu- 
nity and a problem. The opportunity lies in the ability to achieve more 
economic procedures and more effective management planning and 
control. The problem, of course, is to select the electronic system 
which will best accomplish this objective. 

There is no pattern of experience for the selection of an electronic 
system. A review of current installations discloses that the following 
were among the primary reasons for the choice of a specific electronic 
computer: 

1. Availability of equipment. Equipment has been in short supply. 
Some of the earlier installations have been premised upon early 
delivery. 

2. Satisfactory experience with a particular manufacturer. Cus- 
tomer relationships extending over a period of years have been 
continued. Most office-equipment manufacturers are adapting 
their electromechanical systems to the newer electronic systems. 

3. Special assistance in installation. The companies that installed 
the earlier models of various computers received the assistance 
of the manufacturer's technical staff for lengthy periods at no 
extra cost. However, this extensive assistance is generally less 
available now that computers are more widely accepted. 

4. Comparison of alternative proposals, with selection of the "best." 
Some companies have asked manufacturers to study and review 
their needs, and to make recommendations. Other companies 
have made studies of their requirements themselves and have 

212 



Role of the Executive in Selection of an Electronic System 213 
requested manufacturers to submit proposals. A number of the 
companies that have attempted this approach have found that 
manufacturers could not understand the specific needs suffi- 
ciently to recommend the best system. In some cases the dif- 
ferent manufacturers' proposals were difficult to compare. Gen- 
erally speaking, manufacturers do not like this method, nor do 
most experts feel it is advisable except in special situations. 

5. Availability of maintenance, service, and stand-by facilities. 
Dependability of operation has been an important consideration. 
In one case a company was influenced by the fact that the manu- 
facturer's service center was directly across the street. 

6. Special design characteristics of available computers. Some 
companies desire particular devices for checking accuracy of 
data as they are processed. Other companies have required spe- 
cific input and output equipment which was available from only 
one supplier. 

7. Improvement of suitability of the electronic system. A few 
companies and industry groups have engaged manufacturers to 
build equipment specially designed for their needs. In most 
instances the customers believe that the resulting design will 
be more useful to their industry than are the available general- 
and special-purpose systems. 

8. Uncertainty as to performance on company operations. In order 
to test the suitability of various systems, some companies have 
deliberately installed several different machines, both as to 
size and make. Others have appraised equipment by coding 
their routines and test-running them on manufacturers' machines. 

9. Desire to gain experience. Many companies have started with 
medium-size machines in order to gain experience before at- 
tempting large-scale installations. Others have begun with 
large machines. In several such instances the companies already 
plan to replace the machines with newer models. 

10. Technical characteristics. Several companies have assessed the 
speeds, storage capacities, and other characteristics of various 
components of each computer available. They have selected 
the one which appeared to offer the most for least cost, without 
worrying too much about the specific applications. 

11. Recommendation of consultants. Many companies have hired 
independent consultants to make an appraisal of their needs 



214 Electronic Computers and Management Control 

and to suggest the best system. In some situations the executives 
have relied almost entirely upon such advice. 

12. Adaptation of available equipment. Companies have adapted 
available scientific computers and systems to some of their busi- 
ness requirements. For example, one financial company in- 
stalled a medium-size general-purpose scientific computer. It 
uses the machine to make analyses and prepare top-level re- 
ports from data already processed by other equipment. 

13. Other. In some instances selection has been predetermined by 
top executives. Other companies have agreed to partially fi- 
nance the development of new equipment. This is particularly 
true of input and output devices. The hope has been to ob- 
tain more suitable and compatible electronic systems. 

A variety of other factors also have had some influence. 

On the other hand, companies have advanced a number of reasons 
for delaying selection, after careful study of available equipment. 
Among the reasons they have given are the following: 

1. Technological vs. economic obsolescence. Firms with advanced 
electromechanical systems have not always been convinced that 
electronic computers, though technologically superior, were suf- 
ficiently more economical to justify a change-over. Presently 
available electronic systems were not able to show decisive cost 
advantages. Some company studies have indicated, however, 
that electronic systems under development would show more of 
an economic advantage. 

A number of these companies also say that if they were start- 
ing a new installation they would use present computers. 

2. Need for detailed system study. Many companies believe that a 
comprehensive systems study should be made before a computer 
is ordered. Until such studies are made, no orders are placed. 
A few such firms have placed letters of intent to purchase com- 
puters in order to get on waiting lists. Some of these have altered 
their original orders in view of developments. 

3. Executive reluctance. Some executives do not feel their com- 
panies can afford to lead in pioneering work. There are many 
conflicting reports about first installations. Until there is more 
evidence of substantial benefits the risk may be too great. The 



Role of the Executive in Selection of an Electronic System 215 

executives do not want to tinker with organizations which are 
not giving trouble. 

4. Desire to profit from the experience of others. A number of 
companies do not want to commit themselves until they can in- 
corporate the lessons learned from the mistakes of early efforts. 

5. Concern over legal and auditing requirements. Uncertainty as 
to the position of certified public accountants and the courts, 
with regard to electronic and magnetic records, has caused some 
hesitation. 

6. Executive belief that electronics are not feasible. Because of the 
expense of installations, and limitations of input, output, or 
storage, some executives are convinced that electronic systems 
are not suitable for their type of business. 

This diversity of experience indicates that no pattern for selection 
of electronic systems has yet evolved. Companies have found it dif- 
ficult to reach a decision as to their own program. There is no simple, 
rule-of-thumb answer to the problem of selecting an electronic system. 

Many complex decisions must be made. Some are technical. Some 
are administrative. Personnel experienced in all aspects of this situa- 
tion are rare. There are experts in various special areas, but there 
is a problem of communication. 

Lack of understanding of each other's problems is not a new ex- 
perience for executives and the experts. Problems arise for reasons 
of semantics, differences in the frame of reference with which each 
approaches his problems, the variety of backgrounds and training, 
unwillingness to give each other adequate freedom of action, and a 
tendency to expect tangible results too soon. 

Unfortunately, this gap will not necessarily become narrower with 
time. It may tend to expand unless positive action is taken to pre- 
vent the widening. Development of an integrated business system 
requires an appreciation and understanding of the scientific approach, 
which in turn usually requires some acquaintance with mathematics. 
To communicate, therefore, the executive needs better training in 
the fundamentals of science and in the tools which it uses. 

Scientists and engineers, on the other hand, have to know enough 
about management problems and procedures so that they can explain 
their concepts in terms that the executives can understand. Among 



216 Electronic Computers and Management Control 

the executive problems that the scientists and engineers must consider 
is the fact that executives usually must be able to justify allocation of 
resources by the expectation of fairly current returns. Programs for 
computers, therefore, have had to take into account ways to achieve 
short-run as well as long-run pay-offs. 

In an effort to convince management that such pay-offs were avail- 
able, overselling has sometimes occurred. Lingering suspicions which 
resulted are still hampering some attempts to install electronic systems. 

In dealing with this situation, the executive has two major objectives : 

1. To determine whether his company can make effective use of an 
electronic system. 

2. To make sure that his company selects the system best suited to 
its requirements. 

To determine whether his company can make effective use of elec- 
tronics, the executive's role is primarily one of selection of competent 
personnel, providing the environment so they can make the necessary 
factual studies, and selection of general areas for their evaluation. 
To make sure the selected electronic system fits the company's man- 
agement and operating requirements, the executive's role is to ask dis- 
cerning questions which will indicate to him that all the relevant fac- 
tors have been considered. 

Determination Whether the Company Can Use Electronics. The 
specifications of the various available electronic systems are usually 
stated in technical terms. The size of the internal memory of a com- 
puter may be expressed as a number of "words," with so many char- 
acters or "bits" per word. External storage is usually measured by the 
characteristics of tapes, in reels of so many feet, with recording den- 
sities per inch and reading speeds in characters per second. Output 
capacity of printers is in lines per minute, with a certain number of 
characters per line. The flexibility is largely determined by the num- 
ber and variety of instructions which the programmer can use. 

Translation of these specifications to business capabilities is seldom 
a simple matter. Business needs are expressed in terms of types of 
reports, numbers of documents, accessibility of information in various 
combinations, etc. 

Most companies which are seriously considering, or have installed, 
an electronic system have created an analysis group within the com- 



Role of the Executive in Selection of an Electronic System 217 
pany to handle the technical problems. Decision to establish such a 
group, however, raises a number of questions: 

1. What areas should the group investigate? 

2. What type of personnel is desirable? 

3. Who should be chosen to head the group? 

4. What size should the group be? 

5. How can the company personnel be encouraged to assist the 
group? 

Application Areas. Experience has shown that electronic systems 
usually require several applications in order to be economical. These 
applications have been found in various areas: accounting, inventory, 
production scheduling, and analysis of management problems, etc. 
Frequently the best application on which to start was not originally 
obvious. 

Therefore, all areas of the company should be examined, at least to 
some extent. 

Companies may find that one type of electronic system cannot meet 
all their needs efficiently. They may need a combination of several 
types, or several independent systems. For example, production sched- 
uling and inventory control may require an electronic system which has 
a large random-access memory, but which computes rather slowly— 
a file computer. On the other hand, payroll may require more speed 
but less storage, if the data can be sorted before processing. 

Team of Experts. To study the feasibility of using a computer in 
all possible areas usually requires a team of experts. These may in- 
clude men who understand management planning and control, ac- 
counting, application of the tools of management science, production 
control, sales, personnel, purchasing, electronics, and mathematics. 

Yet at the same time it is advisable to keep the team as small as 
possible. The members of the group must be able to communicate 
with one another. For this reason also it may be desirable to have the 
team work on a limited area at first, in order to develop group under- 
standing of each member's language and procedures. 

To keep the group small it is often possible to select men with experi- 
ence in several fields, who can be encouraged to learn still others. 
In addition, the team can be supplemented by using the representatives 
of manufacturers. These outsiders can supply information concerning 



218 Electronic Computers and Management Control 

the abilities of their systems. They can teach company personnel to 

program and code for the computers. They can also assist the team 

in understanding how to handle problems by showing them other 

installations. 

The company can also use consultants. These groups may have 
wider experience than the company men. Consultants may act as 
impartial judges of manufacturers' products. Often they can supply 
men for temporary assignments when experts are required for special 
problems. Some CPA firms, management consulting firms, research 
institutes, and university faculty members have been active in this field, 
but the supply of competent men is limited. 

Head of Team. Proper selection of the head of the analysis group 
is important. In most cases it is desirable to choose someone who has 
a basic understanding of the company objectives and the operating 
problems it faces. However, he must also have other abilities. Un- 
less he has these other qualifications it may be better to go outside 
the company. 

He must have a feeling for research. He must be someone who can 
build the computer system and continue the analysis of management 
needs over a long period, as well as during the immediate installation. 
He must be able to understand electronics in terms of what the sys- 
tems can do for the company, not merely as technical devices. Yet 
he must be able to understand the technicalities sufficiently as to be 
able to appraise coming improvements in the machines as they relate 
to company systems. 

He must be a person who accepts and encourages orderly change. 

Most important, he must have the confidence of executives. A major 
factor in building and maintaining this confidence will be the ability 
to communicate effectively with all levels of management and of oper- 
ating personnel. 

If he himself is not a technician— a scientist, engineer, mathematician, 
programmer, etc.— he must be able to communicate with each of these 
in his group. (Even if he is a technician, it may be difficult for him 
to communicate with men outside his own field. ) 

Support— Budget. Once selected, the head will need continuing 
support. For example, it is pointless to establish a group unless the 
budget recognizes the fact that salaries for the necessary experts in 
this field tend to run higher than companies have heretofore experi- 
enced (unless they have hired expert chemists and research engineers, 






Role of the Executive in Selection of an Electronic System 219 
for example). Moreover, the group requires adequate allowances to 
attend computer schools and conferences, to visit installations, and 
so on. 

Support— Internal Attitude. A major problem for the executive is 
to create a favorable environment in the company. Without this the 
team cannot operate effectively. The group must be able to under- 
stand the need for certain types of information, and in particular be 
told of the various exceptions which occur so infrequently that only 
experienced personnel know of them. Without cooperation, the group 
will not learn of these exceptions until they occur. By then it may 
cause considerable difficulty to weave means for handling them into 
the system. 

Most firms are assuring their employees that no one will be dis- 
charged because of any innovations. But this is only a beginning. 
The status of many employees depends upon the fact that they are the 
only ones who understand certain routines. Now the need for this 
ability, acquired painfully over many years, is to be eliminated. While 
the elimination of dependence on particular humans may be desirable 
from the viewpoint of the company, it seldom will be welcomed by 
the employee unless he has reason to believe that his job will be en- 
larged. 

As industrial psychologists have learned, mere wage assurance is 
not enough. When the company says to the employee, "If you help 
us obtain our organizational goal, we'll help you attain your personal 
goal," the workers tend to lose interest in their work and do only the 
minimum that seems to be expected of them. 

Real acceptance depends upon group participation in working out 
the changes which are going to affect them. The change then becomes 
something that the group does, rather than something which is done 
to them. This point is of such importance that it will be given further 
consideration in the next chapter. 

Administration of the Group That Selects the System. The execu- 
tive who administers the program for selection of an electronic system 
cannot take time to understand all of the technical details. He should 
not try to make all the decisions. Rather his role is to ask discerning 
questions, to make sure that the study group has made a thorough in- 
vestigation and has logical reasons for each of its recommendations. 

The questions can be grouped into four major areas which the execu- 
tive may consider. These areas are: 



220 Electronic Computers and Management Control 

A. Potential applications. 

B. Selection of appropriate electronic system. 

C. Economic feasibility of various alternatives. 

D. Managerial aspects. 

Because every company has its own problems, the relative emphasis 
given to the following questions will vary. The list can be viewed as a 
basis from which the executive can develop the discerning questions 
most suitable for giving direction to his group, to assure himself that 
the necessary studies have been made. 

A. Potential applications. 

1. Have all areas been examined? (See Chapters 4, 5, 8, 9, and 

io.) 

2. Which areas seem to offer the greatest immediate benefits, and 
why? Which offer benefits over the longer term? 

a. How has the effort of the study group been allocated, as 
between areas in No. 2, and why? 

b. What is the condition of the company's present data-han- 
dling system? Is it up to date? What is required to prepare 
for conversion to an electronic system? (See Chapter 9.) 

c. Are there any areas in the company which should be ana- 
lyzed scientifically? Can such a study provide a basis for 
an electronic system? (See Chapter 6.) 

3. Have any trial applications been tested on a rented computer? 
What were the results? 

a. What has been the experience of other companies? (See 
Chapter 4. ) 

b. Has any other company applied new methods of analysis 
to a study of these applications? Did the results of the 
study affect the selection of an electronic system? (See 
Chapters 6 and 8.) 

B. Selection of appropriate electronic system. 

1. How was the scope of the system determined? (See Chapter 

3.) 

a. What are the advantages and disadvantages of large-scale 
general-purpose systems for this company's situation? 

b. What are the advantages and disadvantages of medium 
size general-purpose systems for this company? 

c. Are there applications for special-purpose systems? 



Role of the Executive in Selection of an Electronic System 221 

d. Where do combinations of a, b, and c appear more suitable, 
at least for the present, and why? 

2. What types of components are required? ( See Chapter 3. ) 

a. What inputs are required? Are they available, or when will 

they be available? 
h. What processing units are required? 

c. What types of storage are required? 

d. What output devices will be necessary? 

e. What components are required for integration, by using 
similar media at various input points? 

3. Which components are available? (See Appendix 3.) 

a. What new components are under development by manu- 
facturers? When available? 

b. Should the company sponsor the development of new com- 
ponents? 

4. What is the basis for choice of equipment suppliers, other than 
cost? 

5. a. What is the record of the supplier's equipment as to depend- 

ability, accuracy, and flexibility? 

b. What services does the manufacturer offer? Training? 
Programming? Maintenance? Other services? What is 
the background of his representatives? 

c. What is the financial status of the supplier? What assur- 
ance is there of continuity in provision of adequate service 
and development of better equipment? 

C. Economic feasibility of various alternatives. 

1. What is the estimated cost of operation of various alternative 
systems? ( See Chapter 5. ) 

a. Did the cost calculation include: 

(1) Equipment installation (air conditioning, space con- 
version, etc.)? 

(2) Programming? 

(3) Training? 

(4) Change of operating procedures, including analysis to 
design and install integrated systems, develop new re- 
ports, etc.? (See Chapters 8 and 9.) 

2. What is the cost of present systems? What would the cost be 
if present systems were made more efficient by using methods 
other than electronics? 






222 Electronic Computers and Management Control 

3. In what ways will savings be achieved? 

a. Clerical costs? Machine costs? Reduced investment in 
assets? More effective management decisions? 

b. What are the major assumptions made in calculating sav- 
ings? 

(1) When must more personnel be hired to prepare pro- 
grams, etc.? When will personnel be shifted from 
other duties? 

(2) When can delivery of the equipment be expected? 
What has been the experience of others? 

(3) What effect would delay in delivery have on saving 
anticipations? What would be the cost effect of delays 
due to difficulties in programming? 

(4) How long will it take to test the machine? To trial-run 
programs? Where can older equipment be dropped? 

( 5 ) In what priority will applications be placed on the new 
system? Why? 

c. What are the costs of relocating and retraining displaced 
personnel? What effect will this have on anticipated sav- 
ings? 

d. What effect will expansion or contraction of the company's 
business have on the potential savings? 

e. Would systems under development, if successful, make the 
chosen equipment economically obsolete? 

D. Managerial aspects. 

1. What organizational changes are required before the elec- 
tronic system can be installed? 

a. What should be the organization for the computer? 

b. What impact will the installation program have on the 
present organization for data processing? For data analy- 
sis? 

c. Were the views of representatives of all functional areas 
considered? 

(1) What arrangement was made to obtain cooperation 
from experts in the operations of the various functional 
areas in the company? 

(2) What were the most significant suggestions made by 
each of these representatives? What advantages did 



Role of the Executive in Selection of an Electronic System 223 
they see in their particular areas? What difficulties? 
What data do they need? What do they want? 

( 3 ) What has been done to orient personnel other than the 
representatives? What are the attitudes of these other 
personnel forces? 

(4) What are the data requirements of top management? 
d. How will the company's organization structure be affected? 

2. In what way will the service or product sold to the company's 
customers be affected? 

3. How will the auditing, legal, tax, and similar requirements be 
affected? Will prior approval be required? 

4. What continuing effort is required for further development of 
the electronic system? 

a. What cooperation should be afforded with other users? 
What industry groups? 

b. What suggestions can be made to manufacturers to improve 
equipment for company purposes? 

c. What use should be made of consultants? 

d. What support should be given to basic research in this field 
by universities, etc.? 

Measurement of Requirements. Choice of the best electronic sys- 
tem calls for more than mere selection of a certain machine design of 
particular characteristics or degree of reliability, etc. The quantita- 
tive measurement of the business information requirements is a funda- 
mental problem. Such requirements must be determined before a 
choice can be made as to which system will perform the work at the 
most economical price. 

Few companies now are in a position to set forth their data require- 
ments in sufficient detail for selection of electronic systems. In addi- 
tion, few large-size and medium-size companies have developed their 
system of management planning and control to the degree where the 
information uses can be integrated with machine capabilities. 

Machine Selection. Machines now available can process certain 
types of business data economically. But special abilities may be 
necessary in specific situations. For example, a company may only 
need to obtain certain information rapidly and on a daily basis. It 
should select a special-purpose electronic system that can process and 



224 Electronic Computers and Management Control 

report data rapidly. A general-purpose computer might be used for 
this if it has sufficiently fast access to large volumes of memory, but 
it might not be economical. Another company may require rapid 
preparation of reports for management action. The electronic system 
selected should provide fast sorting speeds, fast comparisons, and 
sufficient print-out. 

When large amounts of simple computations are necessary, with a 
minimum of requirements for printed outputs (such as in scientific 
work), a computer can be selected on the basis of speed of computa- 
tion, accuracy, and reliability. If large print-out is required, e.g., pay- 
roll checks, billings, etc., then a system can be selected that provides 
printers with sufficient output capacity and speed to perform the job. 
If faster billings will increase the turn of accounts receivable, and 
thereby provide the company with additional working capital, then 
faster and more expensive print-out units may be desirable. 

In some cases the problem may be to have access to the data from 
geographical distances. For example, in a system now operating, a 
central warehouse has a number of district offices order upon it. For- 
merly, by the time requests were mailed and returned the customer 
might become dissatisfied. The central warehouse is replenished from 
a production facility which is many miles away. The electronic system 
installed was one which provided multiple inputs and outputs at each 
of these locations, thus providing the rapid communication required. 

Dependability and Accuracy. Several aspects of dependability and 
accuracy merit consideration. By and large, an electronic machine 
usually is dependable. It rarely errs if it is in proper working order 
and is not called upon to do more than the routine job for which it 
was designed. 

The larger the machine, and the more comprehensive the scope of 
its activities, the greater the danger that there will be trouble if it does 
break down or make a mistake. Not only are more operations affected, 
but usually a large machine is not so easily replaced as a person or a 
smaller machine. 

Experience with electronic machines so far has indicated that the 
machines are sufficiently dependable for everyday use. Certain pre- 
cautions in the way of maintenance and stand-by equipment seem ad- 
visable, but in general this is not a major problem. 

Most executives would like to see a specific machine operate success- 
fully before ordering one for themselves. It is the most satisfactory 



Role of the Executive in Selection of an Electronic System 225 

test of dependability in their eyes. Unfortunately, as already indicated, 
there are difficulties with this approach. The equipment now offered is 
not as good as the kind which engineers believe can be produced some 
day. Yet a company may find itself in an adverse competitive position 
if it waits until other companies experiment with present machines and 
thus get too much of a head start. Moreover, because of the need for 
a special approach to each company's individual problems the equip- 
ment that proves able to work well for one may not be entirely suitable 
for another. 

Several suggestions have already been made to meet the problem 
of dependability of specific equipment. Major manufacturers will no 
doubt have spare machines available— not necessarily locally— so that 
if a machine breaks down, another, perhaps several states away, can 
take over. The data may be fed to the distant machine by wire, or 
tapes flown to it, with small loss of time. Several tests of this type of 
operation have already been conducted, with generally favorable re- 
sults. 

As for possible loss of a whole set of data, the solution proposed is 
to keep separate sets of tapes onto which the situation can be read 
off at intervals. A disaster at the computer would only necessitate a 
backing up to the last read-off point, and starting over again from there, 
reusing the inflow of data from then on. Since these storage tapes can 
be prepared without interrupting the operations of the computer, and 
since they can be reused when the data are no longer relevant, the cost 
will not be substantial. This is not considered to be a major problem. 

The accuracy of electronic equipment is remarkable, considering the 
volume of items processed. In several carefully monitored test runs, 
such as the Bureau of Census, many millions of transactions have been 
handled without undetected error. In other installations errors have 
occurred at the rate of only a few a month. By and large, electronic 
accuracy is an asset, not a problem. 

Efficient Programming. Programming skill is very important. 
Though the machines operate rapidly, if they must repeat a compu- 
tation many times, the total difference in time between two methods 
of accomplishing the same thing (and therefore the cost) may be 
substantial. On the other hand, experience has also shown that the 
law of diminishing returns holds in this area. Obtaining the last bit 
of efficiency is usually the result of considerable human effort, which 
is justified only if there is considerable repetition of the routine. 



226 Electronic Computers and Management Control 

A partial offset to the cost of programming is the fact that a "library" 
of prefabricated, checked routines can be built up for certain purposes. 
In some cases these programs are now being supplied by the manu- 
facturer of the computers. Another development in this direction is 
the emergence of "pseudo" codes. In effect, the computer will be asked 
to do its own coding. The original instructions will be in forms more 
familiar to the ordinary business analyst, and the computer will trans- 
late them and amplify them into the coded program necessary to con- 
form with its particular design characteristics. So far there has been 
some development of this concept, but more seems inevitable if manu- 
facturers are to achieve widespread acceptance of their machines. 

Installation Costs. The factors which should be considered in esti- 
mating the potential costs of any mechanical installation— maintenance 
(often supplied with rental by manufacturers), space requirements, 
staff, supplies, and so on— are typical of costs for an electronic system. 
Of these, the staff is likely to be most important, after the basic cost 
of the machine itself. Good analysts will get salaries well into five 
figures, good programmers almost as much. 

The costs of the preliminary studies and of the initial capital invest- 
ment are not in the range required for a desk calculator or typewriter. 
The less expensive electronic systems cost at least $25,000 to buy. The 
larger systems cost more than that per month to rent. 

Estimation of potential clerical-cost savings that may be provided 
by electronic systems is not an easy task. Because of the state of the 
art, the knowledge of electronic systems required to establish savings 
computations is often difficult to obtain. Savings considered should 
not only be those to be gained by using the machines now available 
but also savings from machines which are being designed and devel- 
oped, since new designs are rapidly becoming available. 

Cost comparisons should take into account the cost of changing pro- 
grams, subroutines, and coding in case of adoption of a still newer 
machine in the future. Manufacturers hope that they can develop an 
automatic conversion device for translation of instructions between 
machines, but these are not yet available. As pointed out in Chapter 
2, extensive programming and coding for use on new equipment is 
expensive. 

Obsolescence. The rate of obsolescence is difficult to determine. 
Some manufacturers of electronic systems believe that a two-year pay- 
off is a sufficient requirement for a successful installation. However, 



Role of the Executive in Selection of an Electronic System 227 
as has been shown, clerical-cost savings may be a minor factor in 
evaluating an electronic system. In such cases obsolescence is less of 
a problem. 

The design of most large-size computers now offered has not changed 
radically in the last few years. Current developments in computer 
laboratories are nearing new technological breakthroughs, particularly 
with regard to certain components. However, installation of these 
components will not always affect all parts of the system, so again the 
risk of obsolescence is somewhat lessened. 

Investment Alternatives, Alternatives available to the businessman 
include consideration of large-scale computers, intermediate-speed 
computers, and special-design electronic systems. All three are pres- 
ently available. The intermediate-speed,- or "medium-size," computer 
is less expensive, but has limited capacity in input or output com- 
ponents, as well as in speed of computing. Generally speaking, the 
larger systems have over ten times the speed and more effective flexi- 
bility than several of the leading medium-size computers. Special 
systems, on the other hand, may approach the speed of large-scale 
computers for their specific operations. Though special systems are 
limited in use, yet in some instances it may be wise to select a special 
electronic system that has a quick pay-off. This choice could give 
the company experience in electronics and also provide time to judge 
the developments which will be available in the future. 

Savings Estimate. An electronic system may create clerical-cost 
savings, but still may not always be desirable. There have been in- 
stances of companies in smaller communities that transferred some 
functions to other localities and as a result have been subject to diffi- 
culties, such as union action. Also, the savings may not be realized 
as soon as estimated. 

Much can be done to prepare for the computer, as was pointed out 
in Chapter 2. A company can start to reduce the clerical-labor pool for 
certain operations through labor attrition, even before the delivery of a 
machine. However, there are certain clerical operations, like billing, 
which have to be kept up to date and cannot be postponed. 

Savings are often available without installing a computer. If the 
data output is not needed, the routine should simply be dropped. It 
should not be placed on a computer merely because it can be done 
faster and cheaper. Input data may be unnecessarily detailed for the 
desired output, in which case a computer is not required if a simple 



228 Electronic Computers and Management Control 

method of gathering the data is available. To use such savings in de- 
termining computer investment decisions may result in a wrong con- 
clusion. 

In other situations, a computer might handle the peak load of sea- 
sonal work, but it might not be economical to have it available the 
year around. A solution may be to send seasonal work to one of the 
computer centers which are being established by various manufac- 
turers. Consideration of these alternatives is essential to determine if 
and when a computer should be purchased or leased. 

In spite of the many alternatives involved, the experience of a num- 
ber of large-size companies has shown that there are usually at least 
one or two clerical applications which can justify the investment in an 
electronic computer. Additional savings can be realized through 
faster management reporting and from the solutions derived from the 
newer mathematical decision tools. 

If a company wishes to secure the advantages of better decision 
making, it is not enough merely to study the application of computers 
to their clerical system. They can also realize profits that are just as 
tangible and often far more substantial, through application of elec- 
tronic systems to the needs for better management planning and con- 
trol. 

Conclusions. The selection and installation of electronic systems 
in a manner which takes full advantage of new developments are 
major undertakings. The technical requirements, both of the elec- 
tronic systems and of the company's needs, make it necessary to use 
a team approach. 

The task of the executive is to obtain the assistance of the proper 
experts and to evaluate their recommendations. For this he requires 
some understanding of the general capabilities of electronic systems, 
and of the requirements for everyday operations and for management. 

The actual selection of a computer is only a part of the problem. 
Too often it has been treated as though this were the major, if not the 
only, executive responsibility. 

To obtain the full benefits available from use of electronics, it is 
desirable that the executive assume more responsibility than has been 
necessary for the installation of previous data-handling and communi- 
cation equipment. Only men with broad understanding of a company 
can assess the over-all impact of the new systems. 



CHAPTER TWELVE 



The Challenge to the Executive 



We cannot turn back the clock, and it would be futile— in fact, disas- 
trous—to attempt to do so. As I said before, the law of life for institu- 
tions, as for living species, is adapt or perish. 

. . . much work lies ahead for you to do. To meet the changing 
conditions of this rapidly moving world, management needs new tools 
for use in the making of policy decisions. The challenge before you is 
to use your imagination and specialized knowledge in devising the new 
tools that are required. 

"The Challenge of Industrial Change" 
An Address by Robert B. Dunlop 
President, Sun Oil Company 

Automation, electronics, new methods of analysis, new concepts of 
planning and control, and the other developments discussed earlier, to- 
gether present a challenge to the business executive. The challenge 
exists both within his company and in the social environment. 

Sufficient applications have been made so that a pattern already has 
begun to emerge. The new tools are relieving the human mind of the 
need to perform routine tasks, wherever policies or methods can be 
fairly well established in clerical and production operations. Com- 
bined with new methods of analysis they are beginning to take over 
some middle-management decisions. 

Many executives still have shown little interest in the subjects of 
electronics, mathematics, automation, and so on. Pressure of regular 
work is such that these executives feel they have little time for con- 
sideration of future developments which may not apply to their com- 
panies. 

This attitude may be justified. However, it seems likely that the 
advent of these new tools and concepts is going to change the nature 

229 



230 Electronic Computers and Management Control 

of competition in many fields. Both large and small firms may have to 
adjust to the new circumstances if they are to survive. In the past a 
major advantage of smaller firms has been their flexibility. They could 
quickly change their service to meet individual customer demands, 
whereas larger firms of necessity often were committed to set lines of 
action. Much of the trend toward decentralization among larger com- 
panies has been traceable to the attempt to achieve some of the flexi- 
bility of smaller companies. 

In the future, however, if integrated systems are achieved, size may 
become an advantage for flexibility too. The larger companies, using 
electronic communication and controls, may be able to move quicker 
than their smaller competitors, and at less cost. 

To achieve such integrated systems, including automation, will be 
expensive in some fields. In fact, though they are far from small by 
some standards, it already seems as though the newly merged "smaller" 
automobile companies may find it increasingly difficult to compete 
with the giants in their industry. General Motors, Ford, and Chrysler 
are thinking in terms of billion-dollar change-overs. 

Recent trends in industry have been toward less use of smaller sup- 
pliers in some lines. With automation the trend may be expected to 
continue. A large manufacturer recently installed a highly mechanized 
shop to build the "bits and pieces" which it formerly bought outside. 
It makes the items to production order. Although the cost of produc- 
ing the items is somewhat higher than buying, the company believes 
that this is more than offset by inventory savings and avoidance of 
parts shortages. 

An alternative that appears to have a chance of countering this trend 
is the development of less expensive electronic systems for data proc- 
essing, and less expensive control systems for automation, which 
smaller firms can use. Certain electronic designers and manufacturers 
are working in that direction. The extent to which potential users will 
encourage such developments remains to be seen. 

Though a pattern of change may be evident, it is also apparent that 
the process has hardly more than begun. Developments are scattered. 
One firm is putting some clerical operations on a computer, another is 
investigating automation, a third is doing both in separate divisions of 
the company, a fourth is doing mathematical programming, and so 
on. Management faces a challenge to fill the gaps between such spe- 



The Challenge to the Executive 231 

cialized efforts, and to relate the process of change to the objectives of 
their individual companies. 

There is a question as to whether the best approach will be on an 
individual company basis or on an interindustry basis. 

Executive Responsibility. However the development may be han- 
dled, it appears certain that decisions on important matters will be 
made by executives, not by the technical staff alone. Executives will 
choose among the various possible programs and will administer devel- 
opments. They will determine the broader aspects of resource alloca- 
tion, staffing, organizing, planning, and controlling. They will ap- 
prove the choice of equipment, after making sure that a detailed analy- 
sis has been performed to establish specific requirements. Executives 
will monitor the process of the change-over, both as to its technical and 
its human-relations aspects. 

Integration of Business Systems. A major problem, which has been 
discussed at length in previous chapters, is to integrate business ob- 
jectives with the systems of automation, electronic data processing, 
and management planning and control. Although the need is recog- 
nized, the problems are far from solved. Knowledge in this area is 
inadequate. Compared to the size of the problem there still is not 
enough work being done. 

Operating managers, whose interests are vitally at stake, usually are 
too busy with day-to-day pressures to have much time to study the 
integration problem. When they try to do this, semantic differences, 
lack of technological training, absence of installations they can visit 
for visual education, and, for some, inability to grasp fundamental 
mathematical relationships have served to frustrate their efforts. 

Relocation and Retraining of Personnel. Personnel is another prob- 
lem. Displacement of workers will occur. Some unions have already 
approved a resolution calling for labor-management-government con- 
trol of automation. The turmoil caused by worker-management re- 
lations during the first industrial revolution serves to underline this 
resolution. 

Some companies hope to solve part of the unemployment problem 
by shifting workers into other operations, after retraining, and by hiring 
larger numbers of maintenance men. They also expect that workers 
who understand the company processes will be needed to help with 
systems studies. In view of the new skills required, to prepare workers 



232 Electronic Computers and Management Control 

for the change will be quite an undertaking. Some workers may be 

absorbed by equipment builders, but automation in electronics is far 

advanced. 

A most serious challenge will be to make sure that workers in auto- 
matic factories and offices continue to feel they are making a real con- 
tribution. This already has been a major problem, even without auto- 
mation, as shown by Mayo's famous experiments at the plant of West- 
ern Electric. 

Administration of Research and Development. Administration of 
the necessary research and development will be still another consid- 
eration. Evaluation of long-range, far-reaching research studies has 
seldom been of major concern for most executives, except in the chem- 
ical, electronic, petroleum and similar industries. Fundamental re- 
search work has been done by the government and universities, for 
the most part. But it seems apparent that continued development in 
the fields of management, automation, and data processing must de- 
pend upon research activities by business itself. The scientific worker 
needs to experiment and observe, to test his theories. While some of 
these tests can be carried on in the simulated offices, the best labora- 
tory will be an actual business operation. 

Some firms have already recognized this. However, by and large, 
few executives have shown that they really understand how to admin- 
ister fundamental research. Many of the office-equipment manufac- 
turers had to acquire smaller electronic companies in order to enter the 
computer field. Much so-called research in this country actually has 
been development engineering, attempts to reduce costs of known 
operations, rather than an attempt to take a fresh approach, a study of 
fundamentals. 

The effectiveness of such long-range research is often questioned by 
businessmen accustomed to monthly statements of progress. Funda- 
mental research can solve problems by bypassing them, with a sudden 
spurt, after long periods of seeming lack of accomplishment. Such 
changes can revolutionize an industry before old-line management 
stops making buggy whips. An example would be in aircraft propul- 
sion, where American leadership in gasoline-engine technology was the 
basis for some complacency until Korean War experience showed that 
the Russians had bypassed the problem with jets. 

Should a company research group be encouraged to start with the 
top, with broad, over-all objectives? Or should it start at the bottom, 



The Challenge to the Executive 233 

with specific bottleneck problems which, if solved, would provide im- 
mediate pay-offs, and at the same time give the company a chance to 
learn by actual experience with new tools and methods? There are a 
fairly large number of companies whose efforts to date show little evi- 
dence that the problem has been properly considered. They have no 
program which encompasses the needs of the firm and the objectives 
which the research and applications are to serve. Nor has there been 
proper scheduling, to determine the specific persons who should com- 
prise the research group and when they should be hired, the necessary 
resources and when they should be placed at the group's disposal, and 
so on. Feedback to various executive levels is often weakest of all. 
When executives are not familiar with the type of work that the re- 
search team is doing, sometimes the team can wander far off the reser- 
vation. To give research considerable freedom, but not license, calls 
for nice executive judgment. Possibly the worst way to control a re- 
search effort is one which nevertheless is frequently used— to require 
overelaborate budget justifications. 

Installation of Electronic System. Executives eventually will make 
the decision whether to install presently available equipment, or to 
wait for the appearance of improved models of electronic systems or 
improved automation. Adding to the difficulty is the fact that most 
manufacturers of equipment have backlogs of orders, so that the time 
factor is given extra significance. Although not likely as a general 
rule, it is possible that in given situations, by the time an order has 
been filled, the installation completed and de-bugged, etc., much of 
the system may be technologically obsolete. There may be resultant 
advantages to competitors who waited. On the other hand, there 
often will be substantial competitive advantages from being first with 
an improvement over present methods, even though the installation 
may later be subject to improvement. 

Developing Interest. The first problem in furthering a new idea 
usually is the difficulty of interesting others in its development. 

Although relatively few companies actually have installed large- 
scale electronic systems or have made substantial use of the other new 
tools of management and production, this does not mean that there is 
little interest in these subjects. The reverse is true. In fact, it can be 
considered as a tribute to the alertness of American management that 
the challenge of the new developments has been so generally recog- 
nized. Almost every major company has at least a few individuals who 



234 Electronic Computers and Management Control 

are actively studying these areas and trying to relate them to their 
particular organization's requirements. Some companies have been 
studying the problem for over five years. 

There has been little pattern, however, as to who in the organization 
the people might happen to be. In some cases the president of the 
company has initiated the studies, having learned of the new concepts 
through his connections with such groups as the board of governors of 
a university or of a research organization. Or he has learned of them 
from other executives. In other companies, interest has started among 
a few young men in such places as the industrial, engineering, account- 
ing, or systems and procedures departments. 

Professional Organizations. Associations with allied interests, such 
as the Controllers Institute, The American Management Association, 
The American Institute of Accountants, the American Accounting As- 
sociation, the National Association of Cost Accountants, the Internal 
Auditors, the Machine Accountants, the Systems and Procedures So- 
ciety, and others, have encouraged their members to study these new 
developments and have pointed out the significance that the develop- 
ments will have for their work. As shown in Chapter 6, several new 
organizations have been founded— Operations Research Society of 
America and the Institute of Management Sciences. Their purpose 
is to enable researchers to exchange information and to promote new 
concepts. The engineering societies also have been very active in 
this work. The material presented by these groups has been given 
close attention by many company executives and has resulted in the 
initiation of activity. 

Informal groups of interested businessmen are meeting on a regular 
basis in a number of cities. Attendance is limited to one or two par- 
ticipants from companies actively engaged in operating or installation 
studies. No minutes are kept, so that discussion of problems will be 
as free as possible. 

An example is the Business Electronics Round Table in New York 
City, brought together by Harold Cauvet of General Foods. 

Full-time Assignment. As the potential of new tools and concepts 
came to be realized, interest has been formalized by assignment of 
personnel to study and to recommend potential applications for the 
company. Gradual acceptance, while it takes time, may be necessary 
for many companies, especially where operating executives are un- 
familiar with electronics and mathematics. In one company, for ex- 



The Challenge to the Executive 235 

ample, a research study was undertaken several years ago by order 
of the president, who had been alerted to some electronic develop- 
ments. The effort was largely wasted. Apparently this happened in 
part at least because the company organization and the outside con- 
sultants did not know how to cooperate effectively at that time. The 
company dropped the project temporarily, but now has revived its 
interest and has a large, active research project staffed with full-time 
company men and hired experts. The relationship with the consultants 
has been renewed. 

On the other hand, some lower-management personnel have become 
interested in these fields but have found that they could not get atten- 
tion from top management. Frequently this has resulted in the move- 
ment of these younger men to other companies. The younger execu- 
tives have attended schools and meetings on their own time, have met 
interested executives from other companies, and have been hired away 
by these other executives. Sometimes this has happened at the least 
hint that the young men were available, a circumstance indicative of 
the intense pressure to find capable men in these new areas. 

Sometimes operating personnel have led the way. In some com- 
panies it has become evident to operating managers that the volume of 
paper work and other detailed activities required to carry on the busi- 
ness was increasing at such a rate that the situation would get out of 
hand in a few years if something were not done about it. For this 
reason these companies, some of them in industries seldom thought of 
as leaders in management research, have been in the forefront of the 
investigation of electronic systems. 

Equipment Manufacturers. Manufacturers of electronic equipment 
have been eager to assist in this process, but the effectiveness of their 
efforts is somewhat difficult to assess. In a few cases the electronic 
data-handling systems apparently have been oversold, and already 
there have been rude awakenings. In other cases executives have 
attended manufacturers' computer programming schools and have 
come away with great respect for the machines but without the ability 
to see how the machines could be used to advantage in their own 
companies. Nevertheless, manufacturers have been the best source 
of information. 

In general, most progress has been made by those companies which 
have assigned staff members full time to research study of the new 
systems and of the company's particular requirements. Executives in 



236 Electronic Computers and Management Control 

these companies usually wish they had undertaken full-time effort at 
an earlier date. 

Obstacles. A number of obstacles still confront the company which 
has organized groups to study potential applications of electronics and 
mathematics. 

Resistance to Change. Resistance to change can be expected in any 
company. The difficulty of rerouting any document, reassigning any 
duty, and regrouping any organizational unit is such that executives 
frequently have refused to install a change that is demonstrably an 
improvement. The benefits are not considered to be worth the trouble 
involved. 

Obviously the impact of applications of electronics or of mathemati- 
cal decision methods will rarely be minor, and may be so great as to 
merit the designation of "revolutionary." Those workers who remain 
will have to learn new skills, accept new responsibilities, make new 
personal adjustments with new associates, and accustom themselves 
to working under new controls. In addition to the workers, executives 
will be more affected than has usually been the case in the past. The 
new electronic systems are expected to reduce the number of workers 
with whom some executives deal. The new methods are expected to 
change many executive activities. It is well recognized that most 
employees, even at the upper executive levels, tend to dislike closer 
controls, even when they do not actively resist them. The management 
which attempts to use new communication and analysis methods as 
a club to force better performance from subordinates is almost certainly 
going to face resentment and other difficulties. 

Performance sometimes has been so much improved by a particular 
installation of automation, or of electronic data processing, or of mathe- 
matical-analysis methods, that it caused resentment in those depart- 
ments which did not receive the installation. This has led to pressures 
from these neglected managers to have similar aids given to them even 
though there was not sufficient basis for expecting similar results in 
their type of operation. 

Extent of Research Required. Research into a particular firm's 
needs has been found to be a slow and difficult task. Investigators 
have to have a passion for detail. Each company has its own market- 
ing structure, its own products and its own methods of making them, 
its own organizational structure, and so on. These have to be under- 
stood and measured to a degree of precision seldom before attempted. 



The Challenge to the Executive 237 

For example, it is only recently that some companies have begun to 
realize the extent to which individual departments seem to have a 
"personality" of their own. Practices that work successfully in one 
department may have a completely negative effect when applied to 
another, even when it is engaged in similar functions. 

Diversity of Attitudes. A major problem for the head of a research 
group has been to reconcile the differences in attitudes between the 
various members of the team, and to interpret these attitudes to line 
management. Contrary to some public superstitions, good research 
workers are not necessarily temperamental, but they sometimes have 
been impatient with a manager's delays while he assures himself of 
the feasibility of a program. 

Equipment Limitations. Moreover, the electronic systems now 
available usually do not have logical systems based on the logical 
patterns of a particular business' needs. Though the machines are 
adaptable enough so that they still can give considerable service and 
therefore justify their cost, careful selection of equipment, with re- 
design of certain components, sometimes has been found necessary 
for optimum performance in both clerical and production operations. 

Cost. The greatest obstacle has not necessarily been the cost. Even 
with ample funds companies have found the above difficulties may 
hinder effective action. Certainly, however, cost is a major considera- 
tion. None of the machines and none of the trained personnel can 
be found on the bargain counter. Since the principle involved in al- 
most all of these applications is the fact that the whole system becomes 
involved, sooner or later, it follows that very inexpensive, small-size 
applications are almost nonexistent or trivial. Computers are not 
cheap, and a very "small" linear programming problem is likely to be 
not much more than a statement of the obvious. 

Significance of Decisions. The decisions that executives make now 
on these matters may well set a course of action for a business for a 
long time to come. The history of our industrial corporations is a 
neglected field, but evidence is beginning to show that certain types 
of decisions are vital for the long-range prosperity of a company. 

In the past, decisions of this vital nature frequently were concerned 
with territorial considerations. For example, John Wanamaker re- 
cently abandoned the store in New York. In describing the event, 
business publications laid a major share of the blame to a decision 
made by executives two generations ago. The management at that 



238 Electronic Computers and Management Control 

time was faced with the choice of expanding where they were or of 

moving farther uptown. They chose to expand in a constricted market 

area. 

It may well be that failure to develop an integrated business sys- 
tem or development of one oriented to the wrong objectives may result 
in such competitive disadvantages that recoupment may be just as diffi- 
cult as it would have been for Wanamaker to move its store into a 
more favorable shopping area. 

Mere correctness of the major decision may not be enough. Execu- 
tive maturity will be needed in handling the host of problems which 
will accompany the change. The problems are not solely internal; 
there are major social consequences which they must consider. Co- 
ordination of company developments with the activities of other groups 
in our society will become increasingly important. 

THE CHALLENGE TO SOCIETY 

The challenge to business leaders is broader than that within their 
immediate companies. Coming changes may alter the entire structure 
of business and social relationships. In recent years there have been 
numerous expressions of the need for businessmen to demonstrate 
statesmanship to meet growing social problems. The advent of new 
automatic systems that can reduce certain work-force needs by a factor 
of 75 per cent obviously will create a new social problem, large 
enough by itself. But in addition to this, the talents required of those 
workers who remain will be altered. The nature of the products, the 
marketing channels, and so on, also may be different. 

Rapid change-over in all or many of these would present a serious 
challenge to our society. Articles and editorials already have focused 
attention on the possibility that large numbers of employees may be 
displaced by automation of the office and of the factory. In general, 
the danger that they describe does exist. But it is not necessary that 
the change-over should be calamitous or even particularly disturbing. 
There are a number of problems which must be solved, but there is 
no disaster which must be accepted. 

At the same time, the difficulties should not be ignored. 

How fast can automation be introduced without disrupting society? 
At what rate can capital equipment be replaced? Present rates vary 



The Challenge to the Executive 239 

from 8 to 15 per cent in different areas. Since it may take 5 to 10 years 
to develop equipment and techniques to the point where they will be 
technologically available and economically feasible in many lines, the 
change need not be abrupt. A speeding up of the rate is rather the 
most likely prospect. Considering all factors, such as the current short- 
age of trained personnel, the high cost, and so on, it appears that 
though automation of the office and factory is certain to come, it will 
not come overnight. This period of delay will give businessmen a 
chance to program the change, so that should a surge of effort in this 
direction become inevitable, plans will have been laid as to the best 
way to cope with the situation. 

Orderly Change. The change-over to new tools and new methods 
can be orderly. The tools and methods themselves can be used to help 
business leaders prepare better long-range programs than most of them 
have ever achieved before. The programs can take into account the 
social impact of decisions, as well as more commonly used estimates 
such as demand for product, and cost expectations. Social factors are 
true profit considerations, not benevolences. History always has shown 
that if society is badly disrupted, no one profits except occasional 
opportunists. 

Companies can gear their installations to maintain a progressive 
economy. For example, diminished hirings need not be permitted to 
affect a community without adequate warning. Workers can be told 
that if they leave, they might have to seek different types of jobs. They 
can be given training in advance. 

Stockholders and the capital market will face new problems of evalu- 
ation of securities. Executives can help them consider the over- all 
impact of new financing ventures. If adequate information is made 
available, more informed financial judgments will be possible. 

Most companies in the forefront of new developments apparently 
believe that their owners can be brought to understand the desirability 
of expenditures for such purposes. So far, companies like Metropolitan 
Life, Burroughs, and General Electric have given considerable pub- 
licity to their efforts, including in some cases a description in their 
annual reports. 

As part of the process of change, businessmen can keep the public 
informed of their plans and activities. The last two decades have 
taught business leaders the value of business statesmanship. They 



240 Electronic Computers and Management Control 

have learned that it is dangerous not to consider social consequences. 
It is also difficult to prosper in a society that does not understand or 
appreciate what they, the business leaders, are trying to do. 

Business Statesmanship. Business statesmanship is a development 
of the last generation. There has been a tremendous expansion of 
health plans, of pensions and other employee benefits, of profit sharing, 
and of attempts to stabilize employment the year round. Increased 
use of professionals, such as accountants and lawyers, has been an- 
other influence. 

Businessmen who believe that free enterprise can continue to de- 
velop the best institutions, the most effective production systems, and 
so on, have the responsibility for seeing that the new machines and 
methods are used to best advantage, at a minimum cost to society. 

In particular, they can take up the burden of research. So far, as 
indicated, the development of new methods and machines has been 
sponsored primarily by the Federal government. If our free institu- 
tions are to survive, it seems logical that business also must be willing 
to undertake basic investigations which will benefit all. 

Companies can do this either directly, in private research groups, 
or by sponsoring projects at universities. At UCLA, Harvard, Carnegie 
Institute of Technology, University of Chicago, and many other insti- 
tutions, research groups will not undertake an investigation unless 
results can be made public. Even a secret operation, such as the Air 
Force's RAND Corporation, eventually permits its workers to publish 
many of the methods and principles which they develop. 

Education for Management. As technology has become more im- 
portant to business firms there has been an increased tendency for 
technicians to become managers. It can be expected that this trend 
will continue. Training in electronics and mathematics should enable 
engineers and scientists to become even more important to the firm of 
the future. However, although individual technicians have been bril- 
liantly successful, the trend poses a general social problem. By and 
large, both by training and early experience, most engineers and scien- 
tists are not always able to deal with human factors, with intangibles, 
and with over-all concepts. The technician is trained to deal with 
fairly exact knowledge, as opposed to opinions, which are taught in 
the humanities and social sciences. 

As long as management must deal with risk situations, with large 
areas of uncertainty, a straight engineering approach will not be ade- 



The Challenge to the Executive 241 

quate. Yet, at the same time, so many tools and methods for better 
decision making are being developed that the intuitive hunch player 
in management should usually be at a serious disadvantage in most 
situations. 

It appears that a new type of training may be required for future 
managers. This poses a serious problem for our educational system. 
There are inadequacies at all levels. Some experts are coming to be- 
lieve that a completely fresh look at our system of education may be 
required. 

In the past few years there have been warnings that Russia is turn- 
ing out twice the number of scientists and engineers that the United 
States is training. But the Russians have gone farther than merely 
to increase the number of their engineers— they have established a 
different process of selection whereby grammar-school students are 
examined, and those students who show ability likely to respond to 
technical education are separated into different groups and given a 
special training. 

Whatever one may think of this practice, it is a partial answer to 
the complaint voiced in American universities that high-school grad- 
uates are mathematically and scientifically illiterate. It is not enough 
for Americans to scorn the Russian system. If the need for technicians 
continues to grow, we shall have to make some effort to encourage 
more students to take this training. 

But mere numbers of scientists or engineers is not the whole 
answer. Nor will greater quantities of students, trained by our busi- 
ness schools in the traditional "art" of management, supply the need 
for executives capable of dealing with the new tools and concepts. 
A new curriculum may be required, one designed to meet the different 
demands that a different type of society imposes. The combination 
of electronics and data processing, of mathematics and management 
decision making, and automation, all together in an integrated system 
signifies nothing less than a bringing together of the social, the life, 
and the physical sciences. This is a new concept, one which is only 
beginning to be recognized. 

Social Consequences. The first industrial revolution changed the 
pattern of human life in many ways. There is reason to suppose the 
changes which will follow the second may be fully as great. 

Most certainly the use of new tools and methods will increase pro- 
ductivity—they already have. To what extent productivity will be 



242 Electronic Computers and Management Control 

increased no one can say for sure, but when installations already 
made, using tools that admittedly are imperfect, can cut the number 
of workers to a quarter or a tenth of those formerly employed and 
yet turn out more and better work, then the increases are likely to 
be spectacular. 

Nor do these developments leave us with a limit to further progress. 
Quite the contrary. In the past each manager had to learn "by 
experience"; the art of management was taught primarily by ap- 
prenticeship. Formal learning was based primarily on routine uses 
of certain tools such as accounting and statistics, and a few gen- 
eralized principles, hesitantly recommended. 

Much of the "art" of management will always remain, presumably, 
but now we have the hope of developing whole areas of descriptive, 
predictive, and decision models. Model building can be taught. 
From these models further progress can be made so that, as in the 
other sciences, one generation can build on the work of its predecessors. 

Education may have to be changed, as the student pursues the life, 
social, and physical sciences. But a more productive economy should 
be able to support students both out of and in the plant for an 
additional period. College education may be as common as high- 
school education is today. Better students may go on to advanced 
work and research as a matter of course. In-plant training can also 
be more extensive than it is. 

The new developments are of immense significance for the so-called 
backward areas of the world. By learning the new techniques and 
installing the new tools these countries may be able to skip a whole 
industrial generation and enter directly into the technology of the 
last half of the twentieth century. If they had been forced to rely 
instead on their cheap manpower, they would always tend to lag a 
step behind their richer neighbors. 

Increased emphasis on the scientific approach will not be confined 
to the management of business. Attempts to achieve more accurate 
observation and measurement already have been made in other fields. 
Economics has seen such advances in this type of work that a new 
term, "econometrics," has been coined to describe the activities. So- 
ciology has a similar term, "sociometrics." Medicine is changing to 
adopt the new attitudes, as are other fields, and this trend should 
continue with accelerated force. 

Problems will persist, of course. It may be difficult to keep people 



The Challenge to the Executive 243 

busy in occupations which maintain their self-respect, when only a 
part of the population can turn out most of the products that are 
required. But human wants appear insatiable, so this may not be 
of too much concern. As indicated earlier, the transition stages should 
be the most difficult. And such problems as the increase in numbers 
of our older, retired workers no longer need be a problem. 

A great boon to humanity will be the ending of the need to perform 
dull, repetitive clerical and mechanical tasks. As the first industrial 
revolution relieved our muscles from tasks too heavy or too wearying 
for them, so the second will relieve our minds from tasks that dull 
and deaden. As Norbert Wiener has said, we shall see the "human 
use of human beings." 

I/ENVOI 

And the Pursuit . . . No possible advance in our way of life is 
worth the loss of our liberty. The new tools and methods will not be an 
unmixed blessing. They will provide a means for a centralized con- 
trol of human activities unlike that ever experienced before. Many 
of the devices in Orwell's 1984 are far from fantastic. Some are already 
here. Misuse of them, unfortunately, is also far from impossible. 

Moreover, neither an enriched life nor liberty can assure happiness. 
In this increasingly complex society, no man can be an island. Men 
are bound to each other by their search for material wealth, their 
need for social contacts, their fear of foreign dangers, their drive for 
scientific achievements, their longing for artistic conceptions. Men 
spend the best hours of their days and the best years of their lives 
for these. 

Yet these are not enough. Without spiritual content, they are 
meaningless. 

Men can achieve much, in a material and conceptual way, through 
integration of their efforts. But spiritual value each man will achieve 
alone. 



Bibliography 



In the following bibliography the references are placed under four 
major headings: 

A. Background Reading for Management Planning. 

B. Electronic Systems and Their Applications. 

C. Management Sciences— Operations Research. 

D. Automation. 

In most cases the authors cited have published a number of other 
articles or books. References to these will usually be found in the 
publications listed here. 

A. BACKGROUND READING FOR MANAGEMENT PLANNING 

1. Books 

Barnard, Chester I., The Functions of the Executive, Cambridge: Harvard 

University Press, 1938. 
Dale, Ernest, Planning and Designing the Company Organization Structure, 

New York: American Management Association, 1952. 
Dean, Joel, Managerial Economics, New York: Prentice-Hall, Inc., 1951. 
Drucker, Peter, The Practice of Management, New York: Harper & Brothers, 

1954. 
Fayol, Henri, Industrial and General Administration (English translation), 

London: Sir Isaac Pitman & Sons, Ltd., 1930. 
Gardner, Fred V., Profit Management and Control, New York: McGraw-Hill 

Book Company, Inc., 1955. 
Goetz, Billy E., Management Planning and Control, New York: McGraw-Hill 

Book Company, Inc., 1949. 
Holden, Fish, and Smith, Top Management Planning and Organization, 

Stanford: Stanford University Press, 1948. 
Katona, George, Psychological Analysis of Business Behavior, New York: 

McGraw-Hill Book Company, Inc., 1951. 
Koontz and O'Donnell, Principles of Management, New York: McGraw-Hill 

Book Company, Inc., 1954. 

245 



246 Electronic Computers and Management Control 

Mayo, Elton, Social Problems of an Industrial Civilization, Cambridge: 
Harvard University Press, 1945. 

Parker, Mary Follett, Dynamic Administration, New York: Harper & Broth- 
ers, 1941. 

Simon, Kozmetsky, Guetzkow, and Tyndell, Centralization vs. Decentraliza- 
tion, New York: Controllership Foundation, 1954. 

Taylor, Frederick W., Scientific Management, New York: Harper & Brothers, 
reprinted 1947. 

2. Magazines 

Advanced Management, Society for the Advancement of Management, 74 

5th Ave., New York 11, N.Y. 
Business Week, McGraw-Hill Publishing Company, Inc., 330 W. 42nd St., 

New York 36, N.Y. 
Controller, Controllers' Institute, 2 Park Ave., New York 16, N.Y. 
Fortune, Time Inc., 540 N. Michigan, Chicago 11, 111. 
Harvard Business Review, Soldiers Field, Boston 63, Mass. 
Journal of Accountancy, 270 Madison Ave., New York 16, N.Y. 
Management Review, American Management Association, 1515 Broadway, 

Times Square, New York 36, N.Y. 

3. Other Sources 

American Management Association— series on management. 
Controllership Foundation, Management Planning and Control, An Anno- 
tated Bibliography, Controllers' Institute, 2 Park Ave., New York 16, 

N.Y. 

4. Selected References 

Hurni, Melvin L., "Decision Making in the Age of Automation," Harvard 
Business Review, vol. 33, no. 5, September-October, 1955. 

Luce, Henry R., "A Speculation About A.D. 1980," Fortune, December, 
1955. 

Rothschild and Kircher, "Projecting Capital Needs," Journal of Accountancy, 
September, 1955. 



B. ELECTRONIC SYSTEMS AND THEIR APPLICATIONS 

1. Books 

General Books 

American Management Association, "Electronic Data Processing in In- 
dustry," also "A New Approach to Office Mechanization; Integrated 
Data-processing through Common-language Machines," New York: 
American Management Association, 1955. 



Bibliography 247 

Canning, Richard, Electronic Data Processing for Business and Industry, 
New York: John Wiley & Sons, Inc., 1956. 

Haskins and Sells, Electronic Data Processing, New York, 1955. 

Jacobsen, First Conference on Training Personnel for the Computing Ma- 
chine Field, Detroit, Michigan: Wayne University, 1954. 

Technical Books 

Engineering Research Associates, High-speed Computing Devices, New 
York: McGraw-Hill Book Company, Inc., 1950. 

Hartree, Douglas R., Calculating Instruments and Machines, Urbana: Uni- 
versity of Illinois Press, 1949. 

Richards, R. K., Arithmetic Operations in Digital Computers, New York: 
D. Van Nostrand Company, Inc., 1955. 

2. Magazines 

Automatic Control, Reinhold Publishing Corporation, 400 Park Ave., N.Y. 

Computers and Automation, Berkeley Associates, 36 W. 11th St., New 
York 11, N.Y. 

Control Engineering, McGraw-Hill Publishing Company, Inc., 330 W. 42nd 
St., New York 36, N.Y. 

Data Processing Digest, Canning, Sisson Associates, 914 So. Robertson Blvd., 
Los Angeles 35, Calif. 

Journal of the Association for Computing Machinery, Association for Com- 
puting Machinery, 2 E. 63rd St., New York 21, N.Y. 

3. Other Reference Sources 

All manufacturers publish manuals describing in detail the operation of 
their machines. In addition some equipment manufacturers publish their 
own magazines describing specific applications for their own equipment. 

4. Selected References 

Laubach and Thompson, "Electronic Computers: A Progress Report," 

Harvard Business Review, vol. 33, no. 2. 
Osborne, Roddy F., "GE and the Univac," Harvard Business Review, vol. 32, 

no. 4, 
Proceedings, American Institute of Radio Engineers, October, 1953. 

C. MANAGEMENT SCIENCES-OPERATIONS RESEARCH 

1. Books 

General— None 

Technical 

Bross, Irwin D. J., Design for Decision, New York: The Macmillan Com- 
pany, 1953. 



248 Electronic Computers and Management Control 

Cooper, Charnes, and Henderson, Introduction to Linear Programming, 
New York: John Wiley & Sons, Inc., 1953. 

Duncan, Acheson J., Quality Control and Industrial Statistics, Chicago: 
Richard D. Irwin, Inc., 1952. 

Koopmans, T. C, ed., Activity Analysis of Production and Allocation, 
Cowles Commission for Research in Economics No. 13, New York: 
John Wiley & Sons, Inc., 1951. 

McCloskey and Trefethen, Operations Research for Management, Balti- 
more: The Johns Hopkins Press, 1954. 

McKinsey, J. C. C, Introduction to the Theory of Games, New York: 
McGraw-Hill Book Company, Inc., 1952. 

Morse and Kimball, Methods of Operations Research, Cambridge: Pub- 
lished jointly by the Technology Press of Massachusetts Institute of 
Technology and John Wiley & Sons, Inc., New York, 1951. 

Shannon, Claude, The Mathematical Theory of Communication, Urbana: 
University of Illinois Press, 1949. 

Snedecor, George W., Statistical Methods, Ames: The Iowa State College 
Press, 1950. 

Von Neumann and Morgenstern, Theory of Games and Economic Be- 
havior, Princeton: Princeton University Press, 1944. 

Wald, Abraham, Statistical Decision Functions, New York: John Wiley & 
Sons, Inc., 1954. 

Williams, J. D., The Compleat Strategy st, New York: McGraw-Hill Book 
Company, Inc., 1954. 

2. Magazines 

American Society for Quality Control, 70 E. 45th St., New York 17, N.Y. 

American Sociological Review, New York University, Washington Square, 
New York 3, N.Y. 

Econometrica, University of Chicago, Chicago 37, Illinois. 

Journal of Applied Psychology, American Psychological Association, Inc., 
1515 Massachusetts Ave., Washington 5, D.C. 

Journal of the American Statistical Association, 1108 16th St. N.W., Wash- 
ington 6, D.C. 

Journal of the Operations Research Society of America, Editor, George 
Shortley, Johns Hopkins University, Chevy Chase 15, Md.; Member- 
ship, Norvell E. Miller, Mount Royal & Guilford, Baltimore 2, Md. 

Management Science, Editor, West Churchman, Case Institute of Tech- 
nology, Cleveland, Ohio; Business Manager, Alan Mann, SKF Indus- 
tries, Front and Erie St., Philadelphia, Pa.; for subscriptions or mem- 
bership in the Institute of Management Sciences, Paul Kircher, School 
of Business Administration, UCLA, Los Angeles 24, Calif. 

Society for Industrial and Applied Mathematics, Box 7541, Philadelphia, Pa. 



Bibliography 249 

3. Selected References 

Ackoff, Russell, "Operations Research in Business and Industry," Industrial 
Quality Control, May, 1952. 

Bavelas, A., "Communication Patterns in Task-Oriented Groups," Journal of 
Acoustical Society of America, no. 22, 1950. 

Dantzig, George, "Recent Advances in Linear Programming," Symposium 
on Linear Programming, Bureau of Standards, USAF Controller, Wash- 
ington, D.C., January, 1955. 

Drucker, Peter, "Management Sciences and the Manager," Management 
Science, vol. 1, no. 1., October, 1954. 

Flood, Merrill, "Management Science Today and Tomorrow. Decision Mak- 
ing," Management Science, January, 1955. 

Fortune Magazine: 

"A Theory of Strategy," January, 1949. 
"Operations Research," April, 1951. 
"The Information Theory," December, 1953. 
"How Businessmen Make Decisions," August, 1955. 

Hurni, M. L., "Observations in Operations Research," Operations Research 
Journal, July, 1954. 

Likert, Rensis, "Developing Patterns in Management," General Manage- 
ment Series, Number 178, American Management Association, 1955. 

McGee, John F., "The Effect of Promotional Effort on Sales," Journal of 
Operations Research Society of America, February, 1953. 

Orden, Alex, "Mathematical Solutions of Programming Problems," Manage- 
ment Science, January, 1955. 

Salveson, Melvin, "Research Developments of Quantitative Method in Pro- 
duction," Proceedings 5th Annual Industrial Engineering Institute, Uni- 
versity of California, 1953. 

Schlaifer and Henderson, "Mathematical Programming," Harvard Business 
Review, May -June, 1954. 

Simon and Holt, "The Control of Inventories and Production Rates— A 
Survey," Journal of Operations Research Society of America, August, 
1954. 

Vazsonyi, Andrew, "The Use of Mathematics in Production and Inventory 
Control— I and II," Management Science, October, 1954, and July, 
1955. 

Whitin, T. M., "Inventory Control Research: A Survey," Management 
Science, October, 1954. 

D. AUTOMATION 
1. Books 

General 

Diebold, John, Automation, The Advent of the Automatic Factory, New 
York: D. Van Nostrand Company, 1952. 



250 Electronic Computers and Management Control 

Wiener, Norbert, The Human Use of Human Beings, New York: Hough- 
ton Mifflin Company, 1950. 

2. Magazines 

There are a large number of technical magazines in the electronics and 
general production field which deal with automation. A few of special 
interest: 

Automation, Penton Bldg., Cleveland 13, Ohio. 

Control Engineering, New York, McGraw-Hill Publishing Company, Inc. 

Factory Management and Maintenance, New York, McGraw-Hill Pub- 
lishing Company, Inc. 

Industrial Laboratories, Box 89, Village Station, New York 14, N.Y. 

3. Selected References 

Cf. Control Engineering. Bibliography published serially in various is- 
sues, 1955. 



APPENDIX ONE 

The Language of the Computer 



This section will present a short description of the way alphabetic 
and numeric information, such as is printed on this page, can be 
translated into a code which the computer can use. There is no need 
for the businessman to memorize binary arithmetic, since he will never 
see a report in binary notation. But a knowledge of binary helps one 
to understand how the computers operate. Binary arithmetic has 
been used for machines and systems because of the ease with which 
it can be represented in physical terms, not because it is necessarily 
the best way to count. 

The reader who has no interest in the internal operation of the 
computer can skip rapidly through this section. Those who wish 
to know more than is available in this brief introduction to computers 
should refer to the bibliography. 

The sequence in a counting system can be either 001, 010, Oil, 100 
or 1, 2, 3, 4. Both sets of symbols "say" the same thing. The im- 
portant matter is the treatment of the symbols. For this a set of 
rules is needed, such that the symbol "2" (or "10," in binary) shall be 
equal to the sum of "one" and "one." 

A logical set of rules, useful for machines, was developed by a 
mathematician named George Boole over a century ago. His logical 
system emphasizes use of a minimum of these rules, primarily for 
the concepts "and," "or," and "not." It is usually spoken of as "Boolean 
algebra." 

A simple system of binary counting, with decimal equivalents, ap- 
pears on page 252. 

Each part of the binary number is called a "bit." For example, 1010 
has four bits, of which two are "ones" and two are "zeros." The value of 
this for electronics lies in the fact that the binary number can be 

251 



252 Electronic Computers and Management Control 



Binary 


Decimal 


1 


1 


10 


2 


11 


3 


100 


4 


101 


5 


110 


6 


111 


7 


1000 


8 


1001 


9 


1010 


10 


1011 


11 


etc. 


etc. 



simulated easily by electronic means. The presence or amplitude of a 
charge at a given time, or in a given address, may be considered as a 
"1"; the absence, or different amplitude, as a "0." 

The problem for the design engineers was to achieve this representa- 
tion of symbols in physical terms and to find ways of combining them 
which corresponded to the logical rules of arithmetic. The first rep- 
resentation of binary notation in machines was accomplished with re- 
lay switches. If a switch— like a light switch— was "off," that repre- 
sented "zero"; if "on" that represented "one." A combination of two 
switches could represent "00," "01," "10," and "11" (with both switches 
off, one on, then the other on, then both on ) . 

Addition could be accomplished by connecting the switches so that 
the first electric impulse turned the first switch on, the next impulse 
turned that one off, but in so doing sent on a pulse that turned the 
next one on. For example, see Figure 7. 

Subsequent engineering developed faster methods of representing 
the data, by eliminating these mechanical movements. This was 
accomplishd through a combination of several developments in tubes, 
circuits and diodes, of which a few can be illustrated rather simply. 
One of these was the two-stage vacuum-tube trigger circuit, usually 
called a "flip-flop," a combination of two tubes (or two parts of one 
tube) arranged so that if one tube was transmitting current, it would 
send part of that current to the grid in the other tube in such a way 
that it inhibited current flow in the other tube. Thus if one tube went 
"on," the other would be turned "off." The system was so arranged 
that when a pulse of a certain nature came in, it would turn on the 
tube that was off; this would send a current to inhibit the first tube 



• • 



The Language of the Computer 

SWITCHES \ \ 

BINARY O O 



253 

DECIMAL 



AFTER ONE PULSE 



• • 



• • 



SWITCHES 



B. 



\> 4 

BINARY O I 



AFTER TWO PULSES 



• • 



• • 



C. 



SWITCHES 



BINARY 



4 ^ 

I o 



AFTER THREE PULSES 



SWITCHES 



• • 



• • 



4 4 



BINARY | | 3 

Figure 7 

so as to turn it off. Thus there were two physical stages, both of 
which could not exist at the same time. If the first tube was on, the 
combination indicated a "0," if the second tube was on, a "1." The 
tube combination could represent either "0" or "1," but not both at 
the same time. Since this was an electronic circuit the change from 
"0" to "1" could be made literally in a millionth of a second. 

Also needed was a method of performing arithmetical operations. 
This was accomplished by finding circuit and tube setups in which 
concepts such as the "and," "or," and "not" of Boolean algebra could 
be represented. 

For example, if a tube could only be turned on by a current coming 
from both of two sources, neither of which could turn it on by itself, 
then there is a representation of "and" ( see Figure 8 ) . In this circuit, 
jc, y, and z, notice that you would not want a current from x to circle 




254 Electronic Computers and Management Control 

back into y, but only to go on into z. To prevent this is a function 
of the "diode." There are many places in a computer where the 
designers do not want the current to circle back. Since a diode will 
pass a substantial current in only one direction, putting them on both 

A PULSE COMES OUT ON Z ONLY IF 
PULSES COME IN ON BOTH X AND Y. 

Figure 8 

the x and y lines prevents current from circling back. Diodes may 

be represented: -»- which means the current can go only in the 

direction of the arrow; in this case the current can go to the right, 
but not to the left. 

Now suppose the designer wants a circuit that can add information 
simultaneously from three sources. This requires a more complicated 
circuit; one which uses the "and," the "or," and the "not" of Boolean 
algebra. 

Suppose there are binary numbers "pulses" on three lines, X, Y, and 
Z. If A and B are first cleared and if a pulse comes in on either X, Y, 
or Z, then the result on the counters A and B should be "01." If there 
are pulses on any two of X, Y, and Z, then "10." All three should 
create an "11" situation, since that represents three in binary. 

The circuit shown in Figure 9 will accomplish the desired result, 
using only "and," "or," and "not." (This diagram is of course greatly 
simplified. For example, there would also have to be wires to carry 
the "clock pulse," a timed pulse that in effect "opens the gates" at 
regular intervals so the other stored pulses can each advance one stage 
through the circuitry.) 

In the diagram, 

An "and" box passes a pulse only when a pulse from each of the wires 

is present. 
An "or" box passes a pulse when a pulse from any of the entering 

wires is present. 
A "not" box inverts a pulse. In a sense this creates a pulse if none 

is present. 

Now if you trace the wires you will find that each of the possibilities 
can be handled with the proper result. For example, one pulse on 



The Language of the Computer 



255 



OR 



AND 



AND 



OR 
'MIDDLE' 



AND 



AND AND 



OR 



NOT 



•o" 



-CURRENT GOES 
BOTH TO 
"A" AND "NOT" 



Figure 9 

X, Y, or Z above will get through the OR on the left, be joined by a 
pulse from the NOT through the "middle" OR and thus turn on B, A 
remaining off since no pulse can pass any of the AND boxes on the 
right. If X and Y pulse, then A will pulse, but not B, since neither 
wire into the middle OR will pulse. If all three pulse, both A and B 
will receive pulses. 

These possibilities can be shown in the following table, where a "1" 
below the letter indicates a current came through: 









Binary Count 


Decimal 


X 


Y 


z 


A 


B 


Equivalent 


























1 





1 


1 





1 








1 


1 


1 











1 


1 





1 


1 


1 





2 


1 





1 


1 





2 


1 


1 





1 





2 



Multiplication can be achieved through repeated addition, and di- 
vision through repeated subtraction. Subtraction in binary can be 
handled just as in decimal, through use of complements. 



256 Electronic Computers and Management Control 

For example, an ordinary desk adder often has the "nines" comple- 
ments in small print on the keys, to aid subtraction: 



To subtract 32 from a number, you add 999967 + 1. 



In binary: 



or 



105 
-32 


105 
99967 




72 
1 


73 

10010 
-101 


73 

10010 
+ 11111010 

1 


1101 


0001101 


11011 
-1001 


11011 
+1110110 

1 



Note: The complement 
of these numbers can be 
obtained by reversing all 
positions — all the "0 V 
become "IV and all the 
"IV become zeros. This 
type of maneuver simpli- 
fies the problem of cir- 
cuiting. 



10010 



0010010 



This illustration and discussion have given only an indication of the 
ways in which the circuits can operate. Many varieties of circuits have 
been suggested and tested. A "twos" complement is often used for 
subtraction. Some machines operate in the "parallel" fashion, in which 
all the parts of a number are operated on at the same time; others 
are "serial," in which case the number is worked on a bit at a time. 

Combinations of circuits to achieve such operations as multiplication 
and division become quite complicated. Besides these, circuits are 
needed to carry the instructions, other circuits to find data in storage 
or to put it there, clock circuits are required so that the process will be 
synchronized, and so on. But while the circuitry of the machine is 
complex, control of the operations requires relatively simple com- 
mands to control the operations. For example, general-purpose com- 
puters can perform addition, subtraction, division, multiplication, com- 
parison, shift, transfer, transfer if negative, and so on, as separate 
individual commands, each requiring only a few code numbers. The 



The Language of the Computer 257 

total number of such operations varies between computers, most of 
the general-purpose machines having from 25 to 100 possible opera- 
tions. Each operation has its own coded signal, usually in the form 
of a number. Fox example, on one machine a "14" tells the processor 
to divide, a "19" tells it to multiply, and an "01" tells the computer to 
stop. 

If the sequence of the control signals is built into the computer, it 
becomes a special-purpose computer that can perform only the 
calculations for a particular type of problem. 

Not only are there various systems for handling the data but there 
are also various methods of setting up the data to be handled. Some 
early computers only worked on numbers. Now computers can handle 
letters of the alphabet as well, coding them into binary notation just 
as was done with the decimal digits. Of course, this means that 
some binary numbers have to be reserved for letters, so the number 
notation cannot run as simply as the one illustrated earlier. 

In addition, there are advantages to systems which only use parts 
of the binary numbers that are available. For example, on some ma- 
chines the numbers run like this: 



Binary Coded 
Decimal 


Straight 
Binary 


Decimal 


Binary Excess 
Three 


0000 0000 
0000 0001 
0000 0010 


0000 
0001 
0010 



1 
2 


0000 0011 
0000 0100 
0000 0101 



0000 0011 
0000 0100 


0011 
0100 


3 

4 


0000 
0000 


0110 
0111 


0000 0101 
0000 0110 


0101 
0110 


5 
6 


0000 
0000 


1000 
1001 


0000 0111 
0000 1000 


0111 
1000 


7 
8 


0000 
0000 


1010 
1011 



0000 


1001 


1001 


9 


0000 


1100 


0001 


0000 


1010 


10 


0000 


1101 



Systems that can handle letters of the alphabet and punctuation 
become more complicated still. However, this does not concern the 
user, since business computers automatically "translate" the data from 
numbers and the alphabet to binary, and back again. 



APPENDIX TWO 



Programming 



Computers operate at tremendous speeds and perform their opera- 
tions automatically. But how does a computer know which operation 
to do next? How does it know that it should add one amount to 
another, and then place the total on a certain storage position, and 
so on? 

This is the function of the "program." A program is a set of coded 
"instructions." Each instruction specifies the "command," or "order," 
which tells the machine what to do, and the "address," which tells it 
where to get the data on which to work, and where to put the results. 

There are various types of machines. Some of them, called 
"multiple address," combine several addresses and a command in a 
single instruction. In such cases a single instruction might tell the 
computer where to get each of two numbers, to multiply them together, 
and where to put the result. Some systems only use one at a time 
(in which case it may require several instructions to accomplish the 
same result as the "multiple-address" machine accomplishes with one 
instruction). Each type has advantages and disadvantages. 

When the instruction has been "coded" (given its symbol), it is 
ready for use on the machine. Different machines have different 
codes. For example, one computer is built to interpret "01" as a com- 
mand to stop. In another computer "01" is a command to read input. 

These orders and commands are given in the form of characters. 
Since any particular number might otherwise represent a total of 
items in inventory, or other quantity, this means that the person pro- 
gramming the machine must make certain that the logic of the machine 
will interpret the number as a command, not as a quantity, if that is 
what he wishes to do. Machines vary in logical design, so a pro- 

258 



Programming 259 

grammer has to learn the system of the machine he is to use in order 
to accomplish this. 

The address instruction, like the command, also is in the form of 
number or code. This address identifies the location of the informa- 
tion, just as the address of your home identifies it. When the com- 
puter is to take a number from a specific place, do something with it, 
and then put the number some place else, the computer must be 
given both a coded command and address. Part of the instruction must 
tell the computer where the quantity to be handled is to be found, 
and the other part must tell the computer what operation to perform 
with the quantity. 

The fact that both commands and addresses are in the form of 
numbers is of great importance to the "programmer" of the computer. 
It means that he can change a command or address simply by adding 
or subtracting a number from it. The change can be stored in the 
machine. This helps enable a programmer to make the machine re- 
peat operations, using part of a program over and over. 

In many machines, the instruction is coded in the form of a so-called 
"word." The "word" is a number of a certain length, such as 006-042. 
The six digits are used in writing the program; they will form a 
"word" of about 24 bits when translated into binary for or by the 
machine. The machine is instructed to start at a certain address, 
say, "001," where it finds this word, "006-042." It interprets the word 
as a command. On the next cycle (machines usually operate in 
"cycles"; that is, a pulse of current is created by a clock signal at 
regular intervals, and an operation is or can be performed at each 
new clock signal), the machine will go to address "042," take the 
number "x" it finds at address "042," interpret the number "x" as a 
quantity, and perform the coded command "006" upon it. Then the 
machine goes on, often to the next address (002), in sequence after 
the address where it found the first instruction, interprets the number 
it finds stored in the next address as a new instruction— and so the 
operation continues. 

To program an application such as payroll would be an impossible 
task if each person's pay had to have an individual program. A 
system is required which will enable the machines to perform a long 
series of operations, numbering thousands or millions of steps, by 
repeating those parts of the operation which are the same for each 
employee. 



260 Electronic Computers and Management Control 

A chain of steps is attained through mathematical manipulation of 
the commands and addresses. The machine can repeat a series of 
operations, each group of operations being exactly the same as the 
previous one, except that one new element is introduced each time. 
This new element may be the payroll information about the next em- 
ployee, taken from the next position on an input tape. Or, as happens 
in many engineering calculations, it may be the result of the previous 
operation. The program is designed so the last instruction in the 
sequence sends the computer back to the first instruction, so it goes 
through the same series again. 

The machine can count its operations as it performs them, and can 
use the count to change the address being used by increasing it "one" 
(or any other number) after each operation. Thus the machine can 
be instructed to look for its first instruction in a numbered storage 
cell, say in the position in storage given the "address"— 2011. After 
completing the instruction found there, the machine can add "1" to the 
"2011," obtaining "2012." It will then go to storage location "2012" for 
its next instruction. This may continue until it reaches "3000," at 
which point the instruction in "3000" may be, go back to "2011" again, 
so the cycle repeats. 

In the meantime, the instructions have caused the machine to read 
the next item on the input tape. This means that the whole cycle 
of instructions stored in addresses from 2011 to 3000 will be repeated 
again, using one different input item. Obviously this process, called 
a "loop," can be repeated as many times as desirable. 

The machine can count the number of times the loop is repeated, 
and turn to some other routine after the count reaches a predetermined 
total. Suppose you wish a machine to repeat a calculation 20 times. 
If you put a "19" in the storage, and ask the machine to subtract the 
count of the number of times it has performed the calculation from 
19, and to do this after each calculation, then after the twentieth 
repetition the difference between the two amounts will be negative 
for the first time. The computer can be instructed to start on some- 
thing else at this point by a "transfer if negative" command. 

An example of such a program, with explanations, for a machine 
that has a "three-address" system is shown in the following illustra- 
tion: 

Suppose the machine reads an instruction composed of a "com- 
mand" (first column, below) and three different addresses (next three 



Programming 261 

columns). In the illustration letters are used to show commands 
for easier study; actually, the command also would be given to 
the machine in the form of a binary number. For example, instead 
of Mu. 001 101 104, the first instruction would be of the form 
06 001 101 104. The machine is built so that the electrical im- 
pulses symbolizing "06" cause multiplication to take place between 
the figures stored in storage location 001 and storage location 101. 
The result is stored in 104. 

The first step is to punch cards or prepare tape to show, say, the 
balance of 100 bank accounts. Then the cards or tape are read into 
the machine. The computer can place each separate balance in a 
separate storage, addressed from 1 to 100 in the illustrated program. 

Then cards or tape must be punched for the numbers which convey 
the instructions. The set of these instructions is the "program." These 
cards or tape are then read into the machine, which is set to store 
constants in storage locations 101 through 105, and instructions in 
110 through 114. (Note: 104 is left blank as an operating space.) 

Now the computer is ready. An operator signals it to start. It 
goes to storage location 110. When the machine reads the contents 
of 110, it finds it is to multiply the contents of 001 by the contents of 
101 and put the result in 104. Then, since this was a "multiply" com- 
mand, the machine automatically goes to location "111" ( 1 more than 
"110") for its next command (see explanation column), then to "112," 
then to "113." When it reaches "114," however, it finds a "compare" 
command. This instruction is interpreted in a special way, which can 
call for either of two possible next instructions: 

a. If the number stored in "105" ( Mu. 101 101 104) is larger than 
the number stored in "110," go back to "110" and repeat the cycle. 
Note, the number in "110" started at Mu. 001 101 104 but it has had 
the number in "102" [+ 001 000 000] added to it each cycle, so the 
amount in "110" gradually nears and finally reaches equality with the 
figure in "105." The number in 111 also is changed each time, so a new 
bank account is worked on each time, or: 

b. If "105" is not greater than "110," go to 115 (114 + 1). 

Note: Storing new information erases old; merely reading does not 
erase. 

A general-purpose computer is flexible and can be instructed to 
perform many types of routines. It can be instructed, through the 
inserted program, as to which routine it is to do next. Given the 



262 Electronic Computers and Management Control 

PROGRAM 

Computer 
Storage First Second Third 



Address Command Address 


Address 


Address 


Explanation 




001 


+ 


000 


721 


4.93 


Account 1 


$7,214.93 


002 


+ 


000 


003 


5.08 


Account 2 


35.08 


099 


+ 


000 


016 


442 


Account 99 


164.42 


100 


+ 


000 


001 


234 


Account 100 


12.34 


101 




050 


000 


000 


Rate 5 per cent. 




102 




001 


000 


000 


Number used to modify contents 
of Address 110. 



103 

104 

105 Mu. 

Start 

110 Mu. 



000 



101 



001 



111 


Add 


104 


001 


001 


112 


Add 


110 


102 


110 



113 Add 



111 



114 Compare 105 



115 Finish 



001 001 Number used to modify contents 

of Address 111. 

Spare register for interest. 

101 104 Special constant to control loop. 



101 104 Multiply account No. 1 in address 

"001" by rate in 101 and store 
interest in 104. Later, use address 
"002," etc. 

Add interest in 104 to old amount 
in 001 and store back in No. 001. 

Add contents of Address 110 to 
Address 102 and put result in 
Address 1 10. This changes the ad- 
dress of the account worked on 
each time. 

103 111 Add contents of Address 111 to 

103 and put result in Address 111. 
This changes the addresses worked 
on by "111." 

110 110 If (105) is greater than (110), go 

to 110 and recirculate once again. 
Note that 110 and 111 now operate 
on the next account because the 
number of the address to which 
they refer has been increased by 1. 
The special constant in 105 is the 
final configuration which instruc- 
tion in Address 1 10 will assume for 
this problem. 

After the last item is compounded 
the constant will no longer be 
greater than the contents of 
Address 110, and the next instruc- 
tion will be taken in order from 
Address 115, which here tells the 
machine to stop. It might tell it 
to start another routine. 



Programming 263 

steps of the routine, it will follow them faithfully. As part of the 
program, it can automatically stop this routine and start another 
when it reaches the end of an input tape, or reaches a certain mathe- 
matical result, and so on. 

Large-scale computers gain their versatility because they can per- 
form a number of very simple operations, perform them in any de- 
sired order and very rapidly. They can be instructed to take a figure 
from one place and to put it in another place, unchanged. In doing 
so they may or may not erase the original record, as desired. They 
can change the number by adding it to another, taken from any speci- 
fied address in the system, then place the result in any desired address, 
either in the arithmetic unit where it can be processed further, or in 
one of the storage locations. 

Computers can compare numbers, and place the larger or smaller 
in specified locations, and so on. These are only examples. Through 
proper combination of the various abilities, a "programmer" can set 
forth a series of instructions which will result in any desired per- 
formance. There are so many of these possibilities that he can make 
the electronic machine perform simple accounting routines, as well 
as high-powered mathematical routines. 

As another approach, visualize a computer program something like 
this: Suppose you have an accounting procedure which is handled 
by a roomful of clerks. You want to go on a vacation. The clerks 
can only follow instructions. If you want to avoid a stoppage while 
you are away, you must write out exactly what they are to do in any 
eventuality— that is, a line-by-line description of each type of report 
they are to receive, what they are to do with the information (add 
which figure to which, etc. ) , and where they are to file or report the 
results. 

If you wrote out all the steps for each possible document and for 
each bit of information, then you would have a program. 

The analogy can be carried a bit further. Before going on the vaca- 
tion, you should leave the office for a few hours to see whether the 
clerks can operate from the instructions— which is equivalent to testing 
a program on the computer before you start to run your regular data 
through it. 

Moreover, if you ever go on vacation again, you may be able to use 
the same set of instructions, with a few changes here and there to 
bring it up to date. Similarly, you can build up a library of stored 



264 Electronic Computers and Management Control 

programs, so that the effort required to set up a program need not 
be completely duplicated for similar problems in future. 

You can build up a library of "subroutines'— pertinent parts of a 
routine— so that a future program may consist largely of making choices 
from a group of these subroutines, each of which has previously been 
used and therefore tested. This will cut down considerably on the 
time it takes to program a new set of operations on the computer. 

There is no single way that the various abilities of the machine 
must be combined in order to make a routine. A skillful programmer 
can make up a shorter series of steps and this saves considerable 
operating time on the computer. He can also program better checks 
against errors of various kinds in the information. 



APPENDIX THREE 



Electronic Data-processing Equipment 



This appendix presents specific details concerning electronic equip- 
ment. Most of the equipment described is commercially available or 
exists at scientific installations. In a few cases of special interest, 
equipment in the laboratory stage will be mentioned to show the 
variety of possible designs; for the most part these will be special 
components. 

In no sense is this a catalogue. The purpose is solely to give a 
picture of the characteristics of typical machines. The material is 
not complete in that some machines will not be described. Moreover, 
since the field is developing rapidly, certain details change. For 
current information it is necessary to contact the manufacturer. 

In order to make the material somewhat more understandable, 
certain arbitrary groupings have been made. For example, the 
BIZMAC, IBM 705, RAYCOM, and UNIVAC are discussed together 
as representative of large-scale general-purpose systems. The speeds, 
type of components, and storages are somewhat similar. However, 
there are major differences, particularly in the logical instructions. 
These are not easy to present in summary fashion, yet they combine 
to make each machine quite distinct. As indicated in the text, a 
number of potential customers have full-scale research projects which 
are attempting to assess the suitability of each of these machines for 
particular applications. 

PART A-LARGE-SCALE GENERAL-PURPOSE SYSTEMS 

Section One— Business Computers. As indicated, there are four 
large-scale general-purpose computers in this country offered for 

265 



266 Electronic Computers and Management Control 

commercial use. (Ferranti of England may have a machine in this 
field, otherwise, foreign computers are usually smaller in scale.) Of 
the four, UNIVAC and the IBM 702-705 series have been in full-scale 
production for some time. The RCA BIZMAC was not offered until 
1955 for 1956 delivery, but a model had been built. The Raytheon- 
designed RAYCOM is the latest entry, and for a while should be 
considered as somewhat less proved in performance. The BIZMAC 
and RAYCOM were not in commercial production at the time this 
appendix was written. 

All are in the million-dollar class. Rentals vary from $20,000 to 
$50,000 a month, depending upon the number of auxiliary units. 

The IBM 705 is an improved version of the 702. The UNIVAC also 
has been redesigned, and the new model is called UNIVAC II. In 
both machines a major change was the replacement of prior high- 
speed memory with the new magnetic-core type, which increased 
speed and reliability. 

Machine Manufacturer 

IBM 705 International Business Machines 

UNIVAC II Sperry Rand 

BIZMAC Radio Corporation of America 

RAYCOM Datamatic 

Input-Output IBM 705 UNIVAC II BIZMAC RAYCOM 

Magnetic tape 

Speed, characters per 

second 15,000 20,000 10,000 37,000 

Number of tape units 

possible By using buffers, virtually unlimited. 

Tape reels, capacity Varies; top length about 2,400 feet; recording density, 

125-200 characters per inch; one reel holds several 
million characters; it takes several minutes to run 
from one end to the other. 

All four systems will handle punched cards. The IBM 705 starts 
with punched cards, which it reads directly or can convert to tape. 
It has a typewriter which prepares punched cards simultaneously with 
typed copy. The UNIVAC has card-to-tape converters, but also has 
tape-preparation typewriters. The BIZMAC also has converters, and 
in addition will handle both five- and seven-column punched-paper 
tape. Machines typically read one tape and write on another, but 
some can read and write on the same tape. 



Arithmetic 


IBM 705 


Add 

Multiply 

Divide 


8,400 
1,250 

212 



Electronic Data-processing Equipment 267 

IBM 705 UNIVAC II BIZMAC 
Printers 

Full 100-130 characters, 
lines— per minute 500 or 600 600 

1000 

BIZMAC offers an electronic photoprinter which reads magnetic 
tape and produces 2,000 characters a second— approximately equal to 
a 120-character-per-line, 1,000-line-a-minute printer. All systems can 
be used with punched-card printers, via the converters. 

Operations per Second 

UNIVAC II BIZMAC 

5,000 2,000 

1,000 200 

500 50 

These speeds are only approximate, and have been rounded for 
convenience. A five-digit number was taken as standard. Some 
machines take longer for larger numbers, others do not (up to a point). 
The UNIVAC, for example, has the same speed as shown above for 
a ten-digit number, while the 705 is slower. "Add" and "subtract" are 
the most rapid commands; "divide" the slowest. RAYCOM speeds 
are comparable. 

Storage. All use magnetic cores for high-speed random access, at 
about ten to twenty millionths of a second. Storage capacity for each 
is about 20,000 characters, except for the BIZMAC (4,000). The ma- 
chines can add several core storage units, but the extra memory is 
expensive and probably not economical for most business applications. 

The IBM 705 and the BIZMAC have magnetic-drum memories for 
medium-speed random access. Capacities are 60,000 characters per 
drum, up to 30 drums, for the 705. Capacity of the BIZMAC drum 
is 30,000 characters. 

The main storage in all systems is on magnetic tape. Tape char- 
acteristics were described in the section on input-output. 

Logic. The logic of the systems varies considerably. Basically, each 
of them can perform any arithmetic or comparison requirement and 
can copy and compare alphabetical information. However, these 
operations are performed by combining a number of instructions, 
which vary from machine to machine. As a result, flexibility and speed 
differ for different requirements. A machine which appeared faster, 
in the table given previously, might not be able to do a complete opera- 



Programmers' 


Mack 


Abbreviation 


Cod 


ADD 


G 


ADM 


6 


RAD 


H 


SUB 


P 


RSU 


Q 


MPY 


V 


DIV 


W 



268 Electronic Computers and Management Control 

tion from input to output as fast as another, because of peculiar re- 
quirements of a business problem. 

It is not feasible to discuss all the aspects of logical design, such 
as the fixed word length of the UNIVAC versus the variable word 
length of the IBM 705. 

To give an indication of the type of instructions required, those for 
one machine will be given— for the IBM 705: 



Arithmetic 
Add from memory to the accumulator, where 

arithmetic takes place 
Add from accumulator to memory (core storage) 
Set accumulator to zero, then add from memory 
Subtract from memory to accumulator 
Set accumulator to zero, then subtract from 

memory 
Multiply 
Divide 

The programmer defines the problem and determines a suitable 
machine procedure for solving it. The coder writes out the instruc- 
tions that solve that problem; for example, RAD 0601, etc. The key 
punch operator, or typewriter operator, prepares the input. The ma- 
chine can translate the "RAD 0601" to the "H 0601" which it actually 
uses in internal processing. It does this by means of a special pro- 
gram. In this way it aids in making its own programming easier. 
When the machine reaches this point in the program, it will go to core 
memory location 0601 and place the contents of that location— which 
may be a letter, a single number, or a symbolic mark such as a #- 
in the central processing unit called the "accumulator." The letter, 
mark, or symbol will then be available in the accumulator, for though 
the item was "added," the accumulator was erased of previous contents 
before it arrived. The number is then available to be added to, or 
multiplied by, another number, etc., or perhaps merely relocated in 
another portion of the high-speed core memory or on the magnetic 
drum. 

There are over forty such symbol or number commands for the 
705, not including variations that deal with the sixteen different ac- 
cumulators, which are individually addressable. Those instructions 
not shown deal with such commands as selecting the input or output 



Electronic Data-processing Equipment 269 

tape, card reader, card punch, or printer; also reading, writing, length- 
ening ( roughly equivalent to adding zeros at end of word or number ) 
shortening, rounding ( differs from shortening in that "5" is first added 
to the position to be dropped), rewind tapes, back-space, compare, 
transfer if high, transfer if equal, unconditional transfer ( note, "trans- 
fer if low" is omitted, but the same result can be achieved by pro- 
gramming "transfer if high," then "transfer if equal," then "uncondi- 
tional transfer" ) , etc. 

Section Two — Scientific Digital Computers. There are a large num- 
ber of computers of different designs but similar in that they are pri- 
marily suitable for high-speed scientific computations. Usually input 
and output are more limited than for the commercial computers. 
Speeds are, in general, as fast or even faster. The NORC, built by 
IBM for the Navy, is about eight times as fast as speeds listed above; 
it can make over 30,000 calculations a second. The LARC, designed 
by Sperry Rand, is even faster. 

Since many of these computers are available for commercial work 
on a rental, part-time basis, and since most are used by universities 
or scientific groups that hold classes or occasionally welcome visitors, 
a partial listing may be of interest. 

The division is geographical, with commercial builders shown at 
their headquarters for production, in most cases. 

United States 
California 
Bendix Aviation, Los Angeles. 

Electro-Data (formerly part of Consolidated Engineering), Pasa- 
dena. 
SWAC and BINAC, University of California, Los Angeles. 
Johnniac, RAND Corporation, Santa Monica. 
J. B. Rea, Los Angeles. 
Litton Industries, Beverly Hills. 
Ramo-Wooldridge Corporation, Los Angeles. 
North American Aviation Corporation, Los Angeles. 
Librascope, Inc., Glendale. 

LARC (in design), University of California, Larchmont. 
District of Columbia 

SEAC, DYSEAC, Bureau of Standards. 

NAREC, Naval Research. 

Logistics Computer, George Washington University. 



270 Electronic Computers and Management Control 

Illinois 

AVIDAC and ORACLE, Argonne National Laboratory. 

ILLIAC, University of Illinois. 
Maryland 

ORDVAC, Army, Aberdeen. 
Massachusetts 

MARK series, Harvard. 

Whirlwind series, M.I.T. 

RAYDAC, Raytheon, Waltham. 
Michigan 

MIDAC, University of Michigan. 

Wayne Computer, Detroit. 
Minnesota 

ERA series, Sperry Rand, St. Paul (also called UNI VAC Scien- 
tific). 
New Jersey 

Bell Computer series, Murray Hill. 

Institute for Advanced Study, Princeton. 
New York 

IBM 701, 704 

Remington Rand UNIVAC Division of Sperry Rand. 
Pennsylvania 

Burroughs, Philadelphia. 

Sperry Rand, Philadelphia. 
Virginia 

NORC, Dahlgren (IBM). 
Wisconsin 

WISC, University of Wisconsin. 
Foreign Countries 
Australia 

C.S.I.R.O. Mark I Commonwealth and Industrial Research Or- 
ganization. 
Belgium 

IRSIA-FNRS Computer, Bell Telephone of Antwerp. 
Canada 

UTEC, University of Toronto. 
England 

ACE National Physical Laboratory, Teddington. 

APE(R)C, APE(X)C, APE(X)C, University of London. 



Electronic Data-processing Equipment 271 

Harwell Computer, Atomic Energy Research, Harwell. 

EDSAC, Cambridge University. 

LEO, a commercial computer, built by the restaurant chain, J. 
Lyons & Co. 

Elliott-NRDC, Elliott Brothers, London. 

Manchester Computer, Ferranti Ltd., for University of Manchester. 

TRE Computer, Telecommunications Research, Great Malvern. 
France 

GAMMA, Compagnie des Machines, BULL, Paris. 

CUBA, Labratoire Central de L' Armament, Paris. 
Germany 

PERM, Technische Hochschule, Munich. 

G 1, G 2, Max Planck Institut, Gottingen. 

MARK IV (similar to Harvard design), Technische Hochschule, 
Darmstadt. Also Rechenautomat IPM. 

Z 4, Z 5, Zuse K. G. Neukirchen (one installed in Switzerland). 
Japan 

TAC, University of Toyko. 

Tokyo MARK I, II, Electrotechnical Laboratory, Tokyo. 
Netherlands 

ARRA, Mathematische Centrum, Amsterdam. 

PTERA, Central Laboratory, Postal and Telecommunications, The 
Hague. 
Sweden 

BARK, BESK, Kunglig Tekniska Hogskola, Stockholm. 
Switzerland 

R4S, Technische Hochschule, Zurich. 
USSR 

No precise data available, but analysis of budget reports and 

technical articles indicates that a large number of computers of 

advanced design are in operation. 

PART B-MEDIUM-SIZE GENERAL-PURPOSE SYSTEMS 

There are a number of medium-size systems in commercial pro- 
duction. The position of this equipment is not easy to assess— in some 
ways it is even more difficult to evaluate than that of the larger 
systems. Most of the machines are adaptations of designs that were 
originated for scientific purposes. Because the systems are more 
limited in scope than those described in the previous section, it has 



272 Electronic Computers and Management Control 

been found more difficult to tie them to existing data-processing sys- 
tems than was hoped at first. However, considerable effort is being 
expended to increase the commercial effectiveness of this type of 
equipment. For example, the IBM 650 originally could only use 
punched cards as input-output. Now it has magnetic tape. Other 
machines only had tape; they are being fitted with card-to-tape con- 
verters. 

Generally speaking, the medium-size machine can be distinguished 
from the larger equipment by the fact that it uses a magnetic drum 
for high-speed random-access memory. This makes the arithmetic 
and logical operating time roughly ten times that of the high-speed 
machines. However, limitations exist at other points, such as at 
input and output, so that in many cases the over-all time for processing 
information may be twenty times or even fifty times that of the larger 
equipment. 

The following machines may be considered in this category although 
they differ considerably in certain aspects of their operations (as will 
be noted in some cases): 



Machine 


Manufacturer 


UNIVAC File Computer 


Sperry Rand 


IBM 650 


International Business Machines 


Datatron 


Electro-Data 


Miniac 


Marchant Calculators 


Monrobot 


Monroe Calculating Machine 


CRC and NCR series 


National Cash Register 


Elecom series 


Underwood 


Magnefile 


Electronics 


Circle 


Hogan 


E 101 


Burroughs 


LOP 30 


Librascope 


Alwac 


Logistics Research 



Other manufacturers interested in this type of machine include 
Telecomputing, International Telemeter, Litton Industries, and Mag- 
navox. 

Prices for these machines vary from several hundred thousand dol- 
lars (or about $6,000 to $10,000 a month rental), for those at the 
top of the list (including auxiliary equipment), down to about $50,000 
for those toward the end. All the larger systems have magnetic-tape 
units; some of the smaller ones use punched-paper tape, cards, or 
Flexowriters and keyboards. If magnetic tape is not used, the over- 



Electronic Data-processing Equipment 273 

all processing time increases to as much as one hundred times that of 
large systems, or even more. 

A major attribute of the UNIVAC File Computer is the ability to 
link as many as 24 input-output units, and as many as 10 magnetic 
drums, to the central processor. From this the machine derives its 
name, since it can reach any one of several hundred thousand short 
records in a fraction of a second. 

The IBM 650 originally had card input-output, at which stage it was 
usually considered to be primarily useful for scientific or semiscientific 
calculations. Redesign added capacity to handle six tape units. Over 
a thousand of this model have been installed or are on order. 

The Datatron has the ability to handle punched-paper tape in addi- 
tion to magnetic tape and punched cards. It is somewhat larger and 
faster in operation than most of the others classed as medium-size. 
The manufacturer uses IBM card readers, punches, and printers. 

The Miniac features input typewriters that prepare magnetic-tape 
'"magazines," which are somewhat comparable to the cartridges used to 
transport movie film in some types of cameras. These magazines can 
be mounted and demounted rapidly on the computer. It also can 
handle paper tape. 

The Monrobot design features flexibility of input-output compo- 
nents—teletype, Flexowriter, punched card, paper tape, magnetic tape. 

NCR equipment (formerly Computer Research) emphasizes point- 
of-sale recording. 

Elecom has an unusually efficient sorter. This ability is lacking 
in most systems; although the machines can sort, the process is com- 
paratively slow. 

Magnefile is a smaller machine with a comparatively large drum 
memory. It has been used primarily to record sales data where im- 
mediate totals are desirable. It can be obtained with keyboard input. 

The other machines also have varying characteristics, which will not 
be detailed here, since the intent is only to provide some indication of 
the variety obtainable. 

PART C-SPECIAL COMPONENTS AND OTHER EQUIPMENT 

When one leaves the field of developed systems and begins to ex- 
amine the variety of special components and devices, it becomes diffi- 
cult to give useful summaries. There are literally hundreds of manu- 



274 Electronic Computers and Management Control 

facturers who are interested in this rapidly expanding new industry. 
In addition, the major manufacturers have active research programs 
which are developing devices which they intend to link to their systems. 

An example is the IBM 305. A series of discs is mounted in a 
manner similar to records in a juke box. A moving head can reach the 
surface of any part of any disc in less than a second. This gives the 
device rapid access to millions of characters. (The discs are similar 
in function to a magnetic drum.) Sperry Rand and others have de- 
signed similar disc memories. 

Another is the print reader. Intelligent Machines, Burroughs, IBM, 
and others have developed print-reading devices which work in limited 
applications. A major effort is under way to improve them. 

A different item is the special-purpose system. An example is Tele- 
register's American Airlines Reservisor. There are many others, rang- 
ing from utility power-grid simulators and digital differential analyzers 
such as the Litton-20 to equipment which verges on other fields such 
as telemetering. 

A number of companies have special components, such as Potter, 
( tape readers, storage units ) , Convair ( photographic printer ) , ( Stand- 
ard Register ( devices to code paper document with metallic spots, and 
convert to punched cards, as well as an electronic "smoke printer" ), 
Burroughs (high-speed printer), etc. 

No mention has been made of devices which are used for automation 
of production. Large-scale systems are not yet available in this field, 
i.e., systems suitable for a variety of industries. Instead, each applica- 
tion uses a number of different components, assembled for the par- 
ticular application. 

A few devices, primarily those for assembling small electrical parts, 
have found several applications. An example is the equipment de- 
veloped by General Mills and United Shoe Machinery. 

It is impossible even to list all the manufacturers and types of devices 
available. Perhaps it will serve to indicate the scope of the field by 
mentioning that electronic data-processing devices and automatic 
controls are a several-billion-dollar industry already— all developed 
since World War II— and that forecasts indicate it should double in 
size within a few more years. 



APPENDIX FOUR 



A Mathematical Model for Integrated Data Systems 



Chapters 4 and 5 discussed the background of the data-processing- 
system problem. Chapter 9 indicated some of the types of research 
that are being and can be done at present. Of the "models" referred 
to in these earlier pages, some relate only to parts of the system— to 
applications which can be made today. This section describes such 
a model in more detail. 

A mathematical model of an integrated data system would be based 
on an analysis of the following major elements: 

1. The events to which the source data refer. 

2. The data flow, from the origination of the record, through 
transmittal and processing, to the end stations where the data 
are used. This would include the informal communication sys- 
tem as well as the formal flow. The form of flow would be 
primarily that of "combining" as described in Chapter 9. 

3. The use of the data in programming, scheduling, and feedback. 
The form of flow would be primarily that of "branching." 

Research groups are working on various aspects of the problem. 
The theory of communication developed by Wiener and Shannon is 
being applied in studies of various information systems, studies which 
include a fairly complete analysis of the organizations involved. The 
statistical decision functions developed by Neymann and Wald, and 
the theory of games of Von Neumann and Morgenstern, represent still 
other approaches. Mathematical programming techniques developed 

* The authors are indebted to Irving Lieberman of Litton Industries for his 
aid in the development of the mathematical models in this appendix. 

275 



276 Electronic Computers and Management Control 

by Koopmans, Dantzig, and Marschak, the psychological studies of 
Bavelis, as well as the input-output approach of Leontieff, are further 
examples of the scientific work which is being done. All these ef- 
forts are in fields which have direct bearing on the business data 
processing and the business decision problems which must be solved 
if we are to achieve integrated business systems. 

Although as yet a truly integrated model cannot be achieved, mathe- 
matical models can be developed which describe the part of the 
system concerned with the basic flow of data. Such models can sym- 
bolize the source points of data origination, the form of information 
transmitted, the channels through which the information flows, and 
the activities which receive the reports. The models can be studied to 
determine quickly who originates data, the volumes of flow, the ex- 
tent to which the flows and storages are duplicated, and so on. The 
models also can be used to show the timeliness of the various levels 
of reports. A change in various parts of the system can be traced 
easily to determine the effects the change will have on the over-all 
system. 

By using the models particular problems can be studied, such as 
determining if a feedback system is correctly related to the schedule, 
so that comparable variance data are provided. 

However, while such a limited model may symbolize data flows and 
reports, it cannot measure the effectiveness of the decisions made by 
the end users of the data. If these could be measured and related di- 
rectly to the data-processing system, they could be incorporated into 
the model, in part at least. In other words, the models being developed 
for programming, scheduling, and feedback then could be integrated 
with the data-processing model. 

The remainder of this appendix gives a description of the way a 
mathematical model can be constructed for an existing system of 
gathering and processing data. It is a descriptive model. The purpose 
of constructing such a model is to enable a business to: 

1. Simplify the manner of gathering and processing data. 

2. Determine the results of changes in the system before actual 
changes are made. 

3. Reduce costs by eliminating duplications. 

4. Speed up the reporting. 



A Mathematical Model for Integrated Data Systems 277 

5. Establish a sound basis for selection of equipment for business 
systems. 

The first step is to define explicitly the terms which will be used in 
the mathematical model. These terms are used to describe the typical 
operations in the system, such as the recording of the data on source 
documents, transmittal and reproduction on intermediate reports, proc- 
essing to summary reports, and delivery to persons in particular busi- 
ness functions. Symbols are assigned to each term. 

The following terms and symbols will be used: 

1. Business functions— "W"— those activities which are grouped ac- 
cording to similar kinds of duties, such as production control, cost 
accounting, sales, and so on. The mathematical analyst usually 
can start with the definition of the functions as he finds them in 
a specific company, though some clarifications will probably be 
necessary. 

2. Information documents— "F"— the forms on which data are re- 
corded. F is subdivided into two major classes: 

a. Source forms— "F s "— documents on which information is re- 
corded for the first time. The information is not obtainable 
by an operation from any other form, either by copying or by 
some computation. 

b. Report forms— "F B "— documents for which the information is 
obtained by performing an operation or a combination of oper- 
ations on data from F s . 

3. The information itself is designated as "C." Two classes of infor- 
mation are generally found: 

a. Identifications— "i"— information which describes or identifies. 
Examples are the date, invoice number, payroll number, de- 
scription of material, department number, and so on. 

b. Quantities— "q"— information which gives a quantitative meas- 
ure, which shows amounts. Examples are hours worked, num- 
ber of pieces produced, value of items transferred, costs, reve- 
nues, ages, and so on. 

4. The time ordering- "t"—o£ information is important. This refers 
to the relative time that information arises, or is made into a re- 
port, not the specific date this occurs. A series of times can be 
represented: t± 9 t 2 , t s , U. 



278 Electronic Computers and Management Control 

5. The informal communication channels— 'I" This symbol desig- 
nates the transfer of items by means other than documents. 

Using these symbols, the mathematical model can be made to reflect 
the relationships between the classes of information, the forms on 
which they appear, and the time order of these forms. Steps in the 
development of the model will be as follows: 

1. Determination of the source data forms F s and the information 
collected on each form, the i's and the q's. To distinguish each 
source form, numbers will be added, F sl , F s2 , and so on. The i's 
and the q's also can be numbered, i ± , i 2 , and so on; q ly q 2 , and so 
on. If the same i appears on two different forms, it will be given 
the same number, since this is one of the ways that duplications 
can be uncovered. The time order of the source form can also be 
shown— F% 

2. Determination of the report forms, F B , showing the sequence of 
data flow from the F s , or from other F R . The sequence can be 
numbered as to different types of forms, F B1 , F B2 , etc., and as to 
different occurrences of the same type, F Blt ±, F B1 , 2 , etc. The time 
of each can be noted— F^ u , F® 2 ,i an< ^ so on - Time has been 
omitted from the models below for simplicity of presentation. 

3. Determination of the reports received by each organization (W). 
The various organizations can be numbered— W±, W 2 , and so on. 
If it is necessary to distinguish the various hierarchical levels in the 
organization, another number can be added to indicate this— Wi, 2 , 
Wi,3, and so on. 

4. Determination of the role of the informal communication system, 
showing the types of i's and q's which are transmitted thereby, 
together with the time order of the transfers. 

The model will take the form of a matrix, since this is a simple way 
to show the relationships between two groups of relevant items. 





Fsi 


Fs2 


h 





1 


k 


1 


1 



This shows that information i± appears on form F s2 , and i 2 appears 
on both F sl and F s2 . 



A Mathematical Model for Integrated Data Systems 279 

It is now possible to build a mathematical model of an integrated 
business-data system using the above notations. The data collected 
within a given company can be recorded in a matrix which discloses, 
first, the source forms, and the classes of information gathered by a 
source form. This source-data-form matrix will be given the notation 
of M . 



Mo 





F A 


F s2 


F s3 


... 


* sm 


Last F s 


First i = i\ 


1 





1 




1 




■ 








1 









k 


1 


1 


1 




1 




- 














Last i = i u 








1 




1 




First q — qi 


1 


1 


1 




1 




qi 


1 















qz 


1 















• 














Last q = q y 



















All of the is and g's must be determined. They must then be exam- 
ined in relation to the source-data forms. Either an item appears on 
a form, or it does not. This can be represented by "0" if the item does 
not appear, and by "1" if it does. The matrix, therefore, will show 
only 0s and Is. 

After the M matrix has been completed, the next step is to construct 
a matrix showing the relationship between the F s forms and the first 
level of report forms, F B1 . This matrix is designated Mi. 



280 
Mi 



Electronic Computers and Management Control 





Frii 


Fr\2 


Fru 




Fri p 


F.i 


1 


1 










F a2 


1 













F aS 








1 




1 


• 












F S m 





1 







1 



The matrix is filled out in the same way as the previous one. If any 
information from an F s is used on an F R , then a "1" appears, if none, a 
"0." Note that the "1" shows the information is available and that some 
of it is used, but not that all the information items from the particular 
F s appear on the F B . 

A matrix for second-level-report forms, showing their relationship 
to first-level forms, is designated M 2 . It appears: 



M 2 





FR21 


Fr22 


FR2S 


... 


FR2T 


Fru 


1 













Fru 


1 


1 


1 




1 


Fru 


1 













• 












Fri p 








1 




1 



A similar use of "0" and "1" is made to show that some information 
is taken from one form for use on another. 



A Mathematical Model for Integrated Data Systems 281 

Matrices for subsequent levels will continue with the same pattern. 



M z 





FR31 


FR32 


FR33 


... 


Fr3w 


FR21 


1 













FR22 





1 







1 


FR23 


1 


1 







1 


• 












F R2 r 













1 



The next level is designated M 3 . 

The order of time is greater for the report forms on each higher 
level, in each higher-level matrix. This is necessary, since the system 
predicates that the reports on the higher levels are using data taken 
from those on lower levels, and thus must succeed them in time. 

If new source data appear on a report from one that was previously 
made up from other sources, then the form should bear a report desig- 
nation showing the appropriate report level, not as starting back on the 
M level. In other words, a form may be mixed in that it may con- 
tain both report and source data. But the ordering requires that the 
report level be established by the level attained as a result of the largest 
number of steps or prior processing for any of the information shown 
on the form. 

Study has shown that report-form levels will usually tend to parallel 
a company's organizational levels. Since there are about three to five 
levels in many typical organizations, this sets the number of matrices 
which will be required to reflect this part of the system. Another 
matrix is required to show the functions using the reports. If this last 
matrix can be produced at the level at which we have just concluded 
our illustrations, the matrix can be designated M 4 , and will appear as 
follows, where W indicates functions: 



282 
M 4 



Electronic Computers and Management Control 





Wi 


w 2 


Ws 




w z 


Frsi 


1 





1 




1 


Fr32 


1 













FR3Z 





1 


1 







• 












FR3w 













1 



Analysis of the Model. An example of the use of the model can now 
be given. Consider any row across any matrix. "0" and "1" will appear 
in various boxes. Each time that a "1" appears this indicates a use of 
the information. Therefore the summation of the row ( addition of all 
the i s that appear ) gives an indication of the extent to which that item 
is used. If the sum is "0," the information is not used; it is redundant. 

Next, consider the columns. These show the number of items that 
appear on a particular report (or, in the last matrix, are used by a 
particular function). The sum of the column, therefore, indicates the 
amount of information used on that report or available to that function. 

The rows and columns of matrix M can be summarized as follows: 











Sum 


of 


Sum of 


*ows 


Sum 


Columns 


Sum 


theV 


s 


the q's 


i\ 


3 


Fsi 


5 


2 




3 


h 


1 


F 8 2 


2 


1 




1 


« 


4 


F 8 z 


5 


4 




1 


i u 


2 


Fsm 


4 


3 




1 


3i 


4 












22 


1 












33 


1 













Qv 



A Mathematical Model for Integrated Data Systems 283 

From the summary it can be seen, for example, that information 
item q y is not being collected, although apparently it was listed as one 
of the regular information items during the initial study. Other rows 
show incidence as high as four times, which indicates the level of use 
of the particular items. 

One of the valuable attributes of matrix mathematics is that the 
matrices can be "multiplied" together. In effect, this means that the 
lower-level matrix can be "merged" with the upper, so as to show the 
items of information which are combined for one report, and then com- 
bined again for the higher-level report. 

Multiplication of items on M± and M 2 gives the following results, 
designated M R i'> 



Mo X Mi = M B1 





Frll 


F r 12 


FrlZ 


Frlp 


»i 


1 


2 


1 


2 


h 








1 


1 


k 


2 


2 


1 


2 


• 










iu 





1 


1 


2 1 


Qi 


2 


2 


1 


2 


Q2 


1 


1 








93 


1 


1 








• 










Qy 















The process can be repeated, multiplying M 2 by M 3 , for M R2 and M 3 
by M 4 for M R3 . The result of multiplying the whole series M X 
Mi X M 2 X M 3 is designated as M B . 

If, in either M R1 , M R2 , M RS , or M R , the number in any "box" is greater 
than one, this shows that the information is being repeated and theo- 



284 



Electronic Computers and Management Control 




Figure 10 



retically is redundant. A measure of the redundancy is given. Another 
way of stating this redundancy is to say that the same i's or q's appear 
on more than one source form, or that a source form appears on more 
than one report form, or a combination of both. 

For example, in a complete matrix M R , it may be shown that infor- 
mation item q\ may be available to function W z as many as 19 times. 
Figure 10 shows how this may occur. For the nonmathematical reader 



A Mathematical Model for Integrated Data Systems 285 

this is shown in flow-chart form rather than in matrix arithmetic. 
Both methods show exactly the same thing. 

Figure 10 on page 284 shows that every F s collected q x \ the informa- 
tion was to a large extent redundant for function W e , as well as for first- 
level reports F R11 , F R12 , and F Rlp ; for second-level reports F R21 , F R23 , 
and F R2r ; and for third-level reports, F R31 and F R3w . 

The model given is for the formal part of the communication net- 
work. As indicated earlier, the informal net often plays a major role 
in determining the effectiveness of the system. For this reason it is 
desirable to make up a model of the informal contacts and communica- 
tions, in so far as these can be obtained. 

A simple indication of such informal links can be given in matrix I 
which shows only that certain functions communicate with others on 
a fairly frequent basis: 





Wi 


w 2 


W z 


... 


W m 


TTi 





1 







1 


W 2 


1 













W z 













1 


• 












Wm 


1 





1 








In the matrix Z, "0" indicates no communication of value, and "1" 
indicates informal communication of some value. 

A more refined version can be developed to show the types of infor- 
mation thus transmitted, the is and the q's which flow along these 
informal channels. In addition, the time order of the information can 
be denoted, with the usual t's and their subscripts. The function which 
originates the information can be coded as a subscript to the i's and q's. 

A matrix M a can be constructed, showing the uses made of infor- 
mation items, if the functions are studied in sufficient detail to provide 
the necessary data. If M a is compared with the availability matrix 
M r , the usefulness of the information items can be analyzed. 

M a would appear: 



286 



Electronic Computers and Management Control 





Wi 


w 2 


W s 




w z 


V 
















V 


1 





1 




1 


•l« 





1 










• 












*jU> 





1 










?!l 





1 










g l2 





1 


1 




1 


«i» 





1 


1 




1 














Q 2 " 


1 





1 




1 



Still another matrix, M d , can be constructed, in form similar to M a 
but based upon the desired or optimum items of information that 
should be provided for each function, as determined by systems studies. 

The value of these matrices does not lie in the fact that they show 
anything which could not be represented in other ways, e.g., by 300- 
foot rolls of paper or voluminous manuals. Rather, the value lies in the 
fact that the analyst has called upon mathematics for presentation of 
information in condensed and manipulable forms. 

Not only does the model enable the analyst to picture the whole of 
a very complicated situation in a relatively simplified form, but the 
model offers other advantages. For example, because the relationships 
are delineated and grouped so precisely, it becomes relatively easy to 
divide the systems-study work among the members of an analysis group. 
Each can be given a different part of the study, with precise knowledge 
as to how their part fits into the whole picture. 

Equally important, because the items are defined clearly and their 
Hows established, it becomes much easier to provide the electronics 



A Mathematical Model for Integrated Data Systems 287 

engineers with the volume, quality, and routing specifications which 
are so hard to obtain for present data-processing systems. It is in 
this direction that improvement in special electronic systems can 
most profitably be sought. 

Another advantage is the experimentation which can be carried 
on with the model, to determine the effects of certain suggested sys- 
tem changes, without disturbing the actual system. This is a scientific 
advantage which is almost impossible to obtain, using present methods. 
In present systems, the direct effects of a systems change can usually 
be foreseen, but the second- and third-order effects, together with over- 
laps and other difficulties, are almost impossible to determine in ad- 
vance. Indeed, they frequently go unrecognized even after the system 
change is put into effect. Efficiency suffers, but the cause is not ascer- 
tained. 

Comparison of information transferred by the informal system, and 
that by the formal net, will give a new picture of the communication 
process. It will be possible to decide the extent to which the informal 
net should be encouraged, the amount of data which should be re- 
peated over the formal net for verification purposes, and so on. 

There are many other advantages afforded by the use of this type 
of model. However, the last we shall mention here is the important 
opportunity which is offered the analyst to incorporate into his sys- 
tem the advances which are taking place in programming, scheduling, 
and feedback, both conceptually and in terms of electronic equipment. 

Uses of the Mathematical Model. There is a very fair chance that 
at this point the nonmathematical reader will say, "The matrices de- 
scribed above don't mean much to me. I don't see how I could use 
them in my own company." 

The best answer is that it is not difficult to experiment with this 
type of model, applying it to real problems on a fairly small scale. 
The feasibility of such an approach can be tested by having someone 
in the systems and procedures department prepare only part of the 
model, e.g., matrix M . For example, in a production firm the matrix 
M could be prepared for the production orders. 

In one case the matrix M was prepared solely for labor source 
documents. The source documents used in this study were as follows: 

1. Daily time card. 

2. Daily labor-realization card. 

3. Labor report required by the foreman from each individual. 



288 Electronic Computers and Management Control 

After the i's and q's were listed for each of the source documents, it 
was evident that the i's on the labor source forms made up two-thirds 
of the information gathered and the q's were only one-third. In other 
words, it was evident, contrary to the executive's original impressions, 
that the amount of quantitative data gathered was much less than 
the amount of identification data. 

Next, an examination of the i's disclosed that there was a great deal 
of duplication in the identification data. For example, the production 
order code contained information as to date, account number, and 
lot number. The accounting codes also contained the account num- 
ber and the lot number, although the order of the digits was different 
than for production control. 

The matrix M also made it quite clear that the daily labor-realization 
card duplicated data on the daily timecard. The duplication existed 
because the production department computed daily labor realizations, 
and the timekeeper needed the daily labor timecards to summarize 
the data for the weekly timecard. Later he sent the weekly card to 
centralized tab for payroll computation and for cost-accounting pur- 
poses as well as for reporting labor back to the foreman and to the 
other levels of management. The foreman's daily source forms du- 
plicated the data on the other two forms except that it added one 
new piece of information. The latter dealt with the particular machine 
used by an individual worker during the production day. 

Upon analysis it was determined that development of an integrated 
code, one which would designate organization, contract, and part 
number, would permit the dropping of about one-third of the digits 
currently used in the i's. 

Counting of the digits required for each of the i's and q's enabled 
the analyst to determine the volume of data needed for labor informa- 
tion, both currently and in storage. The amount of clerical labor, 
etc., necessary to compute the various source documents was easily de- 
terminable. In this particular case it was discovered that the time- 
keeper could be used for preparing the reports used by the foreman, 
which saved about four hours of his time as well as released two filing 
cabinets used to store the daily forms. The timekeeper also sorted the 
daily timecards as he checked them, so that when they reached the 
labor-realization clerks the first sorting was accomplished. Savings in 
paper costs were also realized, as one card was designed to serve for 
all three purposes. 



A Mathematical Model for Integrated Data Systems 289 

Future Developments. The preceding section showed how mathe- 
matical "model" building could be used to determine how a business 
firm's present data system could be integrated, and how duplications, 
unused data, etc., could be eliminated. As indicated earlier, this is but 
the first step in the study of integrated data systems. The next major 
step will be integration of the data for reports and other types of com- 
munication in the feedback concept for programming and scheduling. 

This programming and scheduling concept determines how, when, 
and what to produce, or what service to provide that would ultimately 
result in maximizing the use of the resources of the firm. The feed- 
back to the original program and schedule is important to system in- 
tegration. It is not uncommon to find many companies in which pro- 
grams or schedules have been established, but no provision made 
for follow-up. In other cases the follow-up is so slow that no ef- 
fective action can be taken. The report matrix for programming would 
differ from that used in the business processing system, in that it 
would provide for a more detailed examination of the user who reviews 
the report in a given function W. The reason for this is that in a busi- 
ness firm, as in other organizations, the decisions are made by various 
groups of persons, but the profitability of the firm depends on the joint 
result of all decisions. Furthermore, the profitability of the firm will 
depend not only on the various groups' decisions but also on external 
situations not under their control. 

The model would also need to show the extent to which various or- 
ganizational groups knew about the external situation as well as the 
internal situation, and what they need to know in order to make good 
decisions. In addition, the model should provide for the inclusion of 
local information and special data available to each group. 

Models would have to be available for various types of decisions 
involving production, sales, and finance. These models could be 
viewed as fixed models, used to determine the optimal program or 
schedule. The information required to check the variables in these 
models could be determined. In addition, a requirement showing 
how often these variables should be provided could be built into the 
model. The accuracy requirement of the variables could also be 
determinable from the model. 



Index 



Access time, 14, 22, 25, 40-41 
Accounting, for financial analysis, 144, 
145, 165 
for integrated system, 185-189 
as a model, 133 

score card, attention, and problem 
solving, 176-177 
Accounting applications, 49-86 
Accuracy, 16 

Actuarial applications, 52-53 
Administrative problems of computers, 
97-115 
(See also Executive role) 
Admiral Corporation, 194 
Aeronautics, Bureau of, 90 
Air Force supply, 85-86, 135, 203-206 

(See also RAND Corporation) 
Airlines applications, 81-82 
Alcoa, 181-182 
Altaian's, 78 

American Accounting Association, 234 
American Airlines, 81 
American Institute of Accountants, 234 
American Institute of Internal Auditors, 

234 
American Management Association, 54, 

143, 234 
American Newspaper Publishers As- 
sociation, 206 
American Telephone and Telegraph, 
121 
(See also Telephone application) 
Analog computer, 18, 207-210 
Applications, areas, 49-115, 217 
extension of, 169 
in management science, 128-138 
methods used to choose computer, 
116-120 



Applications, sequence of computer, 71, 
100-101, 111-112 

Arithmetic, 46-47 

Army Ordnance Supply, 88 

Associated Merchandising Corporation, 
79 

Atomic energy, 120 

Auditing (see Certified public account- 
ants; Internal auditing) 

Automation, 2, 120, 124, 138-139, 165, 
190-196 
final step in, 191 
(See also Applications; Installation) 

Automobile company applications, 54, 
59 

Band storage (magnetic drum), 41 
Bank of America, 72-73, 93 
Bank applications, 72-73 
Becker, Charles E., 50 
Billing (see Receivables) 
Binary system, 34 

(See also Language of computer) 
Bits, binary, 34 
"Bits and pieces," 230 
BIZMAC, 70-71 
"Black books," 141-142, 162 
"Black boxes," 164 
Blue Shield, 70-71 
Boole, George, 251 
Bradshaw, T. F., 140 
Break-even charts, 162 
Breakdown (stand-by), 63 
Bridgman, Percy, 173 
British Air Force, 121 
Budgets as models, 134 
Buffers, 38 
Burroughs Corporation, 197, 239 



291 



292 



Electronic Computers and Management Control 



Business Electronics Round Table, 234 
Business statesmanship, 240 

"Cam and gear" mechanization, 190 

Canning, Richard, 91 

Capital equipment replacement, 238- 

239 
Card Programmed Calculators, 34 
Carnap, Rudolph, 173 
Carnegie Institute of Technology, 88, 

135, 178, 240 
Cauvet, Harold, 234 
Cell (storage), 42 
Census, Bureau of, 202 
Center, data-processing, 54 
Central Records, 79-80 
Centralization vs. decentralization, ad- 
ministrative problems, 118-120 

computer system, 55, 88 

decision making, 124 

General Electric, 61, 99-100 

integrated business system, 166-167, 
230-231 

integrated data handling, 176-189 

organizational problems, 145-149 
Certified public accountants, 98, 102, 
112-113, 215 

(See also Internal auditing) 
Challenge, to executive, 229-243 

to society, 238-243 
Channels (on tape), 35 
Chase, Stuart, 173 
Chesapeake and Ohio, 70, 105 
Chicago, University of, 240 
Choice of computer, 212-214, 223-224 
Chrysler Corporation, 54, 145, 230 
Closed-loop feedback, 208 
Coding, 105 

(See also Language of computer) 
Collation ratio, 94 
Combinatorial mathematics, 200 
Combining special- and general-pur- 
pose machines, 83 
Commands (see Program of instruc- 
tions ) 
Committees, use of, 99 
Common-language project, 54, 181 
Commonwealth Edison, 65-67 
Communication, 150-189 

barriers to, 172-176 

formal vs. informal, 172 



Comparison of general-purpose, scien- 
tific, and file computers, 94 
Components, 18, 20 
Computation, 196-207 
Computers, 7-31 

choice of, 212-214, 223-224 

comparison of, 94 

types of, 8-10 

(See also Applications; Installation) 
Consolidated Edison, 71 
Consultants (see Experts) 
Control, computer, 21 

of input, 57 

(See also Planning and control) 
Controllers Institute, 234 
Controllership Foundation, 176 
Conversion, to electronics, 52 

to integrated system, 182-184 

of installed electronics system, 56, 
77, 82-84 

time-consuming, 56-57 
Conversion devices, 209 
Cooley, E. F., 52 
Cost, of automation, 196 

of computers, 4, 11, 29, 41, 59-60, 
77-78, 94, 107-110 

of installation, 60, 226-227 

as obstacle to changes, 237 
Cost accounting, 69-74, 101, 144, 160- 
162 

as a model, 133 
Cowles Commission (Yale), 88, 135 
Credit card, 63-64 

Data-processing center, 2 

Data reduction, 197-198 

Dean, Joel, 162 

Debugging programs (testing), 107- 

109 
Decision models (see Models) 
Defense, Department of, 134-135, 192, 

196 
Dennison Manufacturing, 80 
Department store applications, 76-81 
Dependability, 16, 63, 108-109, 224- 

225 
Descriptive models (see Models) 
Devices, 18, 20 
Diebold, John, 2n. 
Differential equations, 199 
Digital computer, 18-31 



Index 



293 



Diodes, 15 

Disc storage (magnetic drum), 41 

Distributon, 76 

Document handling, 93-94 

(See also Sorting) 
Down time, 16 
Drucker, Peter F., 146 
Dunlop, Robert B., 229 
du Pont, 2, 69-70, 121, 144, 181-182 

Eastman, George, 149 

Eckert, J. P., 43 

Economic model, 209 

Education, 240-242 

Eggleston, W. E., 65 

Election forecast, 201 

Electronic "brains," 7, 10, 31 

Electronic systems, 1, 8, 11, 32, 47-48 

Electrostatic storage, 39 

Elevator dispatch, 199 

Engineering Research Associates, 76 

Engineering schools, 195 

Equipment problems, 108-110 

ERMA, 72-73, 93 

Errors, 107-109 

(See also Dependability) 
Exception, management by, 164-166 
Exceptions, handling of, 61, 66-67, 69 
Executive role, 212-243 

(See also Administrative problems) 
Experts, use of, 98, 119-120, 213-214 

Feasibility studies, 50-52 
Feedback, 99-100, 148-168 
File computer, 25 
File maintenance, 67, 83 
Flexibility, of automation, 194 

of computers, 63, 75, 230-231 
"Flip-flop," 252 
Flow chart, 29, 95, 106 
Ford, Henry, 149 
Ford Foundation, 178 
Ford Motor Co., 2, 59, 145, 193, 197, 

230 
Fortune magazine, 140 
Franklin Life Insurance, 50-51 
Functions of computer, 18 
Functions handled, multiple, 54-55 

Gap between management and techni- 
cal workers, 215 
(See also Team approach) 



General accounting, 69-74, 101 
General Electric, 2, 56-57, 59-62, 88, 

97-106, 121, 197, 239 
General Mills, 193 

General Motors, 144, 197, 199, 230 
General-purpose computers, 8, 16-17 
General routines, 107 

(See also Language of computer) 
Georgia Tech, 79 
Government sponsorship of research, 

121-122, 192 
Grosch, Herbert, 106 
Guetzkow, H., 176 
Guided missiles, 199 

Harder, D. S., 2n. 

Hardware, 32 

Harvard, Graduate School of Business 

Administration, 178, 240 
Harvard Business Review, 137, 140, 

167 
Heads, reading, 35 
Heinz, H. J., 145 
Henderson, A., 167 
Hughes Aircraft, 88 
Human judgment, 195 
Hysteresis loop, 38 

IBM 701 and 704, 88 

IBM 705, 51-52, 71 

"Idle time," 175 

Incremental cost computer, 209 

Industrial TV, 165, 172 

"Information engineering," 187 

Input, 19, 23, 33-37, 60, 110 

( See also Special-purpose computers ) 
Inputs and outputs, planning, 153 
Installation, 4, 12, 50-58, 97-106, 233 
choice of computer, 116-120 
competing interests, 105, 118-120, 

124-125 
reasons for delaying, 214-215 
time estimates, 59-60 
Institute of Management Sciences, 122, 

234 
Instructions (see Program of instruc- 
tions ) 
Instruments, 208 

Insurance applications, 50-56, 70-71 
Integrated business system, 117-118, 
139, 169-189, 230-231 



294 



Electronic Computers and Management Control 



Integrated data processing, 166-167, 
179-189 

Integration, 54, 60-61, 79-80, 97, 105, 
114, 124, 141, 169-189 
final step, 191 

Interest, developing, 233-234 

Internal auditing, 112-113 

International Business Machines Corpo- 
ration, 88 
(See also IBM) 

Inventory applications, 74-86, 101-103 

John Plain Company, 75-78 

Kaiser, 135 
Kaufman s, 79-80 
Kimball Company, 80 
Korzybski, A., 173 
Kozmetsky, G., 176 

Lachman, Harold, 75 

Language of computer, 12, 19, 34, 

251-257 
Large-scale computers, 9 
Legal requirements, 54, 215 
Lever Brothers, 193 
Linear programming, 87, 121, 135, 

167-168, 173 
Litton Industries, 88 
Litton-20, 46 
Lockheed Aircraft, 88, 135 
Logic of computer, 30, 46-47 
Logistics (supply), 134-135 
Loop, 29 
Lyons, Ltd., 101-102 

McCormick, F., 149 

McKelway, St. Clair, 112 

Macy's, 80 

Maddida, 46 

Mail-order applications, 75-77, 84-85 

Magnefile, 78 

Magnetic cores, 15, 38 

Magnetic drum, 41 

Magnetic tapes, 34 

Management Science Project, 91, 135 

(See also UCLA) 
Management sciences, 3, 104, 116-138 
Managerial Economics, 162 
Manufacturers, 235 

list of, 269-272 



Mathematical analysis, 116-138, 167- 

168 
(See also Management sciences) 
Mayo, Elton, 232 
Measurement, 152-153, 195, 223 

(See also Management sciences) 
Media, 18, 33-37 
Medium-size computers, 9 
Metropolitan Life Insurance Company, 

52, 239 
Michigan, University of, 178 
Minimum cost curves, 200 
Misconceptions, 10 
Mitchell, Don, 2, 55, 166 
Models, mathematical, 131-139, 149- 

189 
Modular units, 16 
Monitoring, 210 

(See also Input; Instruments) 
Monsanto Chemical Company, 71 

National Association of Cost Account- 
ants, 234 

National Association of Machine Ac- 
countants, 234 

National Cash Register, 80, 205 

National income, example of mechani- 
cal model, 133 

National Society for Business Budget- 
ing, 146 

National Tube (see United States 
Steel) 

Navy, Bureau of Aeronautics, 200 
Office of Naval Research, 91, 135 
Project Tinkertoy, 196 

New Yorker magazine, 112 

Nonnumeric operations, 208 

Obsolescence, 14 

economic vs. technological, 56, 214 
of special-purpose equipment, 83-84 

Obstacles to change, 236-237 

Ogden, Charles K., 173 

Oil companies applications, 63-64 

Operating characteristics of computers, 
18 

Operation of computer, 23 

Operations research, 3, 119 

(See also Management sciences) 



Index 



295 



Operations Research Society, 122, 234 

Optimizer, 136, 151 

Organization problems of computer 

groups, 104-105 
Organizational planning, 145 
Output, 22, 26, 44-45 

Pacific Mutual, 55-56 

Parallel operation (test runs), 60, 107 

Pay-out time, 55, 82 

Payables (see Receivables) 

Payroll applications, 56-63, 101-103 

Peoples Gas Light and Coke, 67 

Permanence of records, 16 

Personnel problems, 4, 12, 105 

(See also Programmers) 
Plain, John, Company, 75-78 
Planning and control, 2, 114-115, 125, 

139-147 
Point Mugu, 200 
Point-of-sale recorders, 36, 80-81 
Potter Instrument, 80 
Powlison, Keith, 140 
Predictive models (see Models) 
Print readers, 64, 72 
Printers, 44-45 

Procedures improvement, 50, 61-62 
Processing, multipurpose, 59 
Processor, 19, 23, 26, 28, 45-47 
Production control and scheduling, 87- 

93, 103-104 
Productivity, 241-242 
Profit motive, 140 

Program of instructions, 13, 21, 24, 28, 
31, 41, 61, 225-226, 258-264 

time estimates, 97-108 
Programmers, 56, 67, 91, 102, 10S-108 
Programming (planning), 148-168 
Prudential Insurance, 51-52 
Pseudo codes, 107 

(See also Language of computer) 
Public utilities (see Utility company 

applications ) 
Publicity by users, 56 
Punched cards, ability of operators, 34, 
107 

heralded, 125 

(See also Input) 
Punched tape, 182 

(See also Input) 



Quadratic equations, 200 
Questions, discerning, by executive, 
220-223 

Radar, 199 

Radio Corporation of America, 
(BIZMAC), 70, 71 

Railroad applications, 70, 82 

Ramo-Wooldridge, 88 

RAND Corporation (Air Force), 88, 
135, 154, 160, 178, 240 

Random access (see Access time) 

Raytheon (Raydac), 200 

Readers, 64, 72 

Reading heads, 35 

Real time, 198 

Receivables, 50-51, 63-69, 101-103 

Refineries, automation, 195 

Register, 24, 38 

Reliability (see Dependability) 

Remington Rand, 76 

Rentrix, Ltd., 80 

Research, administration of, 232-233 
business sponsorship of, 240 
government sponsorship of, 121-122 
nature of computer installations, 

113-114 
obstacles to change, 236-237 
programming (planning), 153 

Reservisor, 81-82 

Retraining personnel, 231-232 

Return on investment, 144-146, 152 

Richards, Ivor A., 173 

Rich's, Atlanta, 78 

Robinson's, Los Angeles, 80 

Russian developments, 175, 210-211, 
241 

Sales applications, 74-86 
Salveson, M. E., 104 
Scandex, 64 
Scheduling, 148-168 
Schlaifer, R., 167 
School for programmers, 102, 108 
Scientific method, 3, 120, 122-138 
Scientists in business, 121 
Search theory, 137 
Sears, Roebuck, 77, 80 
Seebury Farms, 135 
Selection of computer, 212-214, 22$- 
224 



296 



Electronic Computers and Management Control 



Semantics, 173-176 

Sequence of applications, 71, 100-101, 

111-112 
Serial access, 22 

(See also Access time) 
Service centers, manufacturers, 12, 197 

213 
users, concept of internal, 99 
Servomechanism, 190 
Shannon, Claude, 208 
Simon, H. A., 176 
Simulation, 154, 157-158, 207-210 
SKF Industries, 2, 88, 105, 135, 181- 

182 
Social consequences of change, 4, 

241-242 
Sorting, 65, 109 

of documents, 64, 72 
(See also Access time) 
Special-purpose computers, 9, 72-86, 

92-93 
Specialists (see Experts) 
Specialization in engineering, 195 
Speed, 13, 33, 43 

of data processing, 55 
Speed Tally (John Plain), 75-78 
Sperry Rand, 76 

(See also UNIVAC) 
Spiegel Corporation, 84-85 
Standard Oil of New Jersey, 2, 88 
Stand-by, 63 

(See also Service centers) 
Stanford Research Institute, 72 
Statistics (see Management sciences) 
Steele, Floyd G., 46 
Stock Exchange, 82 
Storage, 22, 27, 34, 37-43 
Strategy, 152-156 
Stryker, Perrin, 140 
Sun Oil Company, 229 
Swift, 149 

Swiss utility model, 209 
Sylvania, 2, 54-55, 166 
Symbols, 171-176 
Systems, business, 30, 139 

detailed description needed, 57, 
192, 214 
(See also Electronic systems; Pro- 
cedures improvement) 
Systems and Procedures Group, 98, 102 
Systems and Procedures Society, 234 



Tabulating cards (see Punched cards) 

Tactics, 152-156 

Tag-omatic, 80 

Taylor, F. W., 121, 125 

Team approach, 3, 30, 60-61, 91-93, 

101, 160-161, 168, 217-218 
Technical training of managers, 5, 242 
Telecomputing Corporation, 80 
Telemetering, 198 
Telephone application, 67-69 
Teleplotter, 200 
Teleregister Corporation, 81 
Test runs, 60, 107 
Testing, equipment, 109 

programs, 107-109 

(See also Dependability) 
Thompson Products, 88 
Tinkertoy, Project, 196 
Top-management support, 98, 218-219 
Toronto Stock Exchange, 82 
Transceiver, 85 

(See also IBM) 
Transistors, 15 
Transportation studies, 121 
Troost, George, 54 
Tyndall, G., 176 

UCLA, 88, 91, 240 

(See also Management Science Proj- 
ect) 
Underwood, 89 
Union attitudes, 231-232 
United Shoe Machinery, 193 
United States Steel, 2, 59, 62, 105, 

135, 181-183 
UNIVAC, 50-57, 61-62, 101, 201-202 
Utility company applications, 64-69, 
71, 209 

Vazsonyi, A., 88 

Wanamaker, John, 237-238 
War games, 207 
Wayne University, 197 
Western Electric, 232 
Wiener, Norbert, 243 
"Wily Wilby," 112 

Yale University, 88 



Date Due 
Due Returned Due Returned 


OCT |6 '56 








NOV \ 1 '56 








MAW* 1 ® 








ADD tm w» 








Urn Qw W 








MAY 71 '5 8 








*0V 3 Jfsfi 








SEP ^| ^ v 








DEC Si W 








■*■'■** 
















jrh| «* 








JAH|2 ^ 








FEBB1 '62 








WAP j*!& JAfl . 








NOV g^'g 








fflgBT^W 








^ffafflfiSat 








WAR^X-i-iW 


1 ^fTJlpTtM 






fiftg * ® wm 








Ml ■■■ ^*^- 


D ;,. ^~^ 






""* ' <- '. 


m * 7» 






AUG It^II 


AUG i * If 


































ENGINEERING 
SCIENCES. 



ilpi 

s 1^62 09364 9928 



* MO